A control method of a quadruped bionic robot applied to planetary exploration
By constructing an environmental cognition model and dynamically adjusting gait parameters, combined with model predictive control and impedance control, the problem of gravity environment mismatch in planetary exploration of quadrupedal bionic robots was solved, and robust planetary exploration capabilities were achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHANDONG FIRST MEDICAL UNIV & SHANDONG ACADEMY OF MEDICAL SCI
- Filing Date
- 2026-04-15
- Publication Date
- 2026-07-03
Smart Images
Figure CN122331418A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of robot control, and in particular relates to a control method for a quadrupedal bionic robot applied to planetary exploration. Background Technology
[0002] Quadrupedal bionic robots are intelligent mobile platforms designed based on bionic principles. They possess advantages such as high load capacity, light weight, and flexible movement, demonstrating excellent terrain adaptability. However, when applied to planetary exploration missions, their bionic structure faces challenges in motion coordination under different gravity environments: the joint torque optimized for Earth's gravity environment is mismatched with the low-gravity environment of planets, leading to decreased gait stability.
[0003] In summary, the performance of existing quadrupedal bionic robots is insufficient to meet the operational requirements when applied to planetary exploration, and improvements are needed. Summary of the Invention
[0004] Therefore, it is necessary to provide a control method for a quadrupedal bionic robot applied to planetary exploration to address the above-mentioned problems.
[0005] The present invention is implemented as follows: a control method for a quadrupedal bionic robot applied to planetary exploration, comprising the following steps:
[0006] By integrating data from multiple sensors to construct a local elevation map of the terrain and simultaneously estimating the robot's posture, and combining it with pre-set planetary gravity parameters, an environmental cognition model is constructed that includes information on environmental geometry (terrain slope, ruggedness) and physical properties (friction coefficient, ground stiffness).
[0007] Based on environmental geometric information (terrain slope, ruggedness), gait parameters (such as duty cycle, stride) are selected or adjusted online from the gait database. Model predictive control (MPC) is used to generate the body's motion trajectory. At the same time, the swing leg trajectory generation algorithm is used to calculate the foot parabola to ensure that obstacles are crossed and the landing is impact-free.
[0008] The whole-body control (WBC) framework unifies the body pose tracking and foot contact constraints into a quadratic programming problem. Impedance control is used to adjust the stiffness and damping of the legs to achieve compliant ground contact. A state machine is used to manage the phase switching of each leg to maintain dynamic balance.
[0009] A hierarchical navigation architecture is adopted; global path planning generates obstacle avoidance topology routes, and local planning updates the landing area in real time based on perception; by evaluating terrain accessibility and energy consumption, the movement behavior (such as detouring and crossing) is dynamically adjusted using reinforcement learning strategies.
[0010] In one embodiment, the present invention provides a control method for a quadrupedal bionic robot applied to planetary exploration. The step of fusing multi-sensor data to construct a local terrain elevation map and simultaneously estimating the robot's posture, and combining preset planetary gravity parameters to construct an environmental cognition model containing environmental geometry and physical characteristics information, specifically includes:
[0011] The robot uses a lidar and stereo vision system to collect three-dimensional point cloud data of the planet's surface. After the three-dimensional point cloud data is processed by filtering and segmentation algorithms, a local elevation map of the planet's surface is generated. Passable areas and obstacles are marked on the local elevation map to establish a passable area map of the planet's surface.
[0012] The robot reads the contact force data from the six-dimensional force sensor on the sole of its foot and the acceleration and angular velocity data from the IMU (Inertial Measurement Unit). The read data is then input into an extended Kalman filter for fusion processing to obtain the robot's precise position and three-dimensional attitude in the planetary coordinate system.
[0013] Based on a map of traversable areas on the planetary surface and the robot's precise position and three-dimensional posture in the planetary coordinate system, combined with pre-set planetary gravity parameters, an environmental cognition model is constructed that includes information on environmental geometry (terrain slope, ruggedness) and physical properties (friction coefficient, ground stiffness).
[0014] In one embodiment, the present invention provides a control method for a quadrupedal bionic robot applied to planetary exploration. The method includes the steps of selecting or adjusting gait parameters online from a gait library based on environmental geometric information, generating the robot's motion trajectory using model predictive control, and simultaneously calculating the foot parabola using a swing leg trajectory generation algorithm to ensure obstacle crossing and impact-free landing. Specifically, these steps include:
[0015] Based on the environmental geometric information of the environmental cognition model, the most suitable basic gait pattern for the current terrain is selected from the gait library, and the duty cycle and stride parameters of the gait are dynamically adjusted by a fuzzy logic controller according to the real-time terrain ruggedness and slope data, so as to achieve adaptive adjustment of gait parameters.
[0016] The model predictive control algorithm is used to continuously optimize the trajectory of the body's center of mass and the position of the zero moment point in the future time domain based on motion commands (such as going to a specific coordinate point) and local elevation map, thereby completing the optimized generation of the body's motion trajectory.
[0017] The system calculates the foot trajectory for the leg in the swing phase, achieves obstacle crossing through cubic spline curve planning, and uses fifth-order polynomial interpolation at the end of the trajectory to ensure that the vertical velocity of the foot returns to zero, generating a safe and smooth foot trajectory to ensure obstacle crossing and impact-free landing.
[0018] In one embodiment, the present invention provides a control method for a quadrupedal bionic robot applied to planetary exploration. The method involves solving the body pose tracking and foot contact constraints as a quadratic programming problem using a whole-body control framework, adjusting the stiffness and damping of the legs using impedance control to achieve compliant ground contact, and maintaining dynamic balance by managing the phase switching of each leg through a state machine. The specific steps include:
[0019] The obtained body motion trajectory, the motion state of each leg and foot end, and the ground reaction force are uniformly modeled as a constrained quadratic programming problem. The optimal torque command for all joints is solved to complete the calculation of the whole-body coordinated control quantity.
[0020] Based on the principle of impedance control, the stiffness parameters of the legs are dynamically adjusted according to the motion phase of each leg: low stiffness is used in the swing phase to ensure flexibility, high damping mode is switched at the moment of ground contact to absorb impact, and impedance parameters are adaptively adjusted according to the physical characteristics information (such as ground stiffness estimation) in the environmental cognition model in the support phase to achieve a smooth interaction between the foot and the planetary surface.
[0021] By managing the switching logic of each leg between swinging, landing, and supporting phases using a finite state machine, and dynamically adjusting the coordination relationship between the four legs based on real-time attitude errors, the robot achieves dynamic balance control on the planetary surface.
[0022] In one embodiment, the present invention provides a control method for a quadrupedal bionic robot applied to planetary exploration, which adopts a hierarchical navigation architecture; global path planning generates an obstacle avoidance topology route, and local planning updates the landing area in real time based on perception; by evaluating terrain drivability and energy consumption, the method dynamically adjusts the movement behavior using a reinforcement learning strategy, specifically including:
[0023] Based on digital elevation maps generated from planetary orbit images, the A* algorithm is used to plan the optimal obstacle avoidance topology path connecting multiple exploration points and generate a global navigation route.
[0024] At the local navigation level, based on the real-time updated environmental cognition model, the safety, slope and slip risk of each potential landing point are assessed to complete the selection of a fine landing point;
[0025] By invoking a reinforcement learning strategy network trained on a large number of planetary environment simulations, and comprehensively analyzing environmental geometric and physical characteristics, motion energy consumption, and scientific task priorities (such as the value weight of the probe target) in the environmental cognition model, specific behavioral strategies are decided online to achieve energy-optimal intelligent exploration decisions.
[0026] In one embodiment, the present invention provides a control method for a quadrupedal biomimetic robot applied to planetary exploration, further comprising:
[0027] The robot uses foot force sensors to monitor the interaction force between its feet and the planetary surface in real time, and immediately initiates a response strategy when a slipping tendency is detected. Based on the principle of impedance control, the robot actively enhances the downward pressure of its feet on the planetary surface by increasing the impedance parameter in the vertical direction of the feet, simulating the effect of pushing off the ground to increase the maximum static friction and enhance ground grip. At the mechanical level, the robot controls the variable foot structure based on the environmental geometry and physical characteristics information in the environmental cognition model. When a smooth rock surface is detected, micro-thorns are deployed, and when loose sand is detected, a structure that increases the ground contact area is deployed. The robot uses mechanical adaptation to provide additional adhesion and improves the traction performance of the robot in the low gravity environment of the planet.
[0028] In one embodiment, the present invention provides a control method for a quadrupedal biomimetic robot applied to planetary exploration, further comprising:
[0029] Using data from the robot's IMU and joint torque sensors, the actual gravitational acceleration and dynamic parameters (such as mass and moment of inertia) of the leg links in the current environment are estimated in real time when the robot is in a stable motion phase. If the estimated values deviate from the preset planetary gravity parameters, the preset planetary gravity parameters are dynamically corrected, and the robot's mass changes caused by carrying or consuming samples are adapted to the changes.
[0030] In one embodiment, the present invention provides a control method for a quadrupedal biomimetic robot applied to planetary exploration, further comprising:
[0031] The robot's posture is continuously monitored by the IMU. When a sudden tilt that exceeds the safety threshold is detected, the preset anti-tipping rescue gait or leg extension and retraction movements are invoked to prevent the robot from tipping over.
[0032] In one embodiment, the present invention provides a control method for a quadrupedal biomimetic robot applied to planetary exploration, further comprising:
[0033] By analyzing data from the robot's leg joints (such as motor current, temperature, encoder feedback, etc.), it is possible to diagnose whether the robot's leg joints are overloaded, stuck, or degraded. If a fault is diagnosed in a particular leg of the robot (such as a decrease in joint torque), the robot's normal gait is switched to a fault-tolerant gait preset in the gait library (such as a triped limp gait) to maintain the robot's basic mobility and operational capabilities.
[0034] In one embodiment, the present invention provides a control method for a quadrupedal biomimetic robot applied to planetary exploration, further comprising:
[0035] The robot's thermal management is achieved by utilizing the diurnal temperature range on the planetary surface. When the ambient temperature is higher than a preset first temperature, the maximum output torque of the motor is limited or a low-heat gait (such as a "crawling" gait) is selected to prevent overheating. When the ambient temperature is lower than a preset second temperature, the robot's internal temperature is maintained by instructing the joints to perform periodic small movements to prevent the lubricant from solidifying and ensure the normal operation of the joints at low temperatures.
[0036] Compared with the prior art, the beneficial effects of the present invention are: the present invention dynamically corrects the planetary gravity parameters, so that the robot motion control matches the actual gravity environment of the planet, thereby ensuring the stability of the core gait; in response to the problem of insufficient foot gripping force caused by low gravity, the present invention monitors the foot sliding trend in real time and adjusts the vertical impedance parameters and foot mechanical shape in conjunction, thus effectively ensuring the robot's passability on complex terrain. Attached Figure Description
[0037] Figure 1 This is a schematic diagram of the first part of a control method for a quadrupedal bionic robot applied to planetary exploration, provided by an embodiment of the present invention.
[0038] Figure 2 This is a schematic diagram of the environmental perception and modeling process provided in an embodiment of the present invention.
[0039] Figure 3 This is a schematic diagram of the gait planning and motion generation process provided in an embodiment of the present invention.
[0040] Figure 4 This is a schematic diagram of the attitude stabilization and force control process provided in an embodiment of the present invention.
[0041] Figure 5 This is a schematic diagram illustrating the navigation and autonomous decision-making process provided in an embodiment of the present invention.
[0042] Figure 6 This is a schematic diagram of the second part of a control method for a quadrupedal bionic robot applied to planetary exploration, provided by an embodiment of the present invention.
[0043] Figure 7 This is a schematic diagram of the third part of a control method for a quadrupedal bionic robot applied to planetary exploration, provided as an embodiment of the present invention.
[0044] Figure 8 This is a schematic diagram of the fourth part of a control method for a quadrupedal bionic robot applied to planetary exploration, provided by an embodiment of the present invention.
[0045] Figure 9 This is a schematic diagram of the fifth part of a control method for a quadrupedal bionic robot applied to planetary exploration, provided as an embodiment of the present invention.
[0046] Figure 10 This is a schematic diagram of the sixth part of a control method for a quadrupedal bionic robot applied to planetary exploration, provided as an embodiment of the present invention. Detailed Implementation
[0047] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.
[0048] It is understood that the terms "first," "second," etc., used in this application may be used herein to describe various elements, but unless otherwise specified, these elements are not limited by these terms. These terms are used only to distinguish one element from another. For example, without departing from the scope of this application, a first script may be referred to as a second script, and similarly, a second script may be referred to as a first script.
[0049] In one embodiment, such as Figure 1 As shown, a control method for a quadrupedal bionic robot applied to planetary exploration includes the following steps:
[0050] Step S1: Integrate data from multiple sensors to construct a local elevation map of the terrain and simultaneously estimate the robot's posture. Combine this with preset planetary gravity parameters to construct an environmental cognition model that includes information on environmental geometry (terrain slope, ruggedness) and physical properties (friction coefficient, ground stiffness).
[0051] Step S2: Based on environmental geometric information (terrain slope, ruggedness), select or adjust gait parameters (such as duty cycle, stride) online from the gait library, use model predictive control (MPC) to generate the body's motion trajectory, and use the swing leg trajectory generation algorithm to calculate the foot parabola to ensure that the body crosses obstacles and lands without impact.
[0052] Step S3: The body pose tracking and foot contact constraints are solved as a quadratic programming problem through the whole body control (WBC) framework. The stiffness and damping of the legs are adjusted by impedance control to achieve compliant ground contact. The phase switching of each leg is managed by a state machine to maintain dynamic balance.
[0053] Step S4: A hierarchical navigation architecture is adopted; global path planning generates obstacle avoidance topology routes, and local planning updates the landing area in real time based on perception; by evaluating terrain accessibility and energy consumption, the movement behavior (such as detouring or crossing) is dynamically adjusted using reinforcement learning strategies.
[0054] In planetary exploration missions, steps S1 to S4 aim to construct an autonomous and robust motion system capable of adapting to complex and unknown terrain. The fundamental objective is to establish an environmental cognition model in real-time, incorporating information on the geometric and physical characteristics of the environment through multi-sensor data fusion, providing a basis for motion decisions. Based on this, gait parameters are adjusted online according to environmental geometric information, and model predictive control is used to generate the body and foot trajectories, ensuring the robot can smoothly and without impact traversing obstacles. Furthermore, through a full-body control framework and impedance adjustment, compliant control of the legs and feet and maintenance of dynamic balance are achieved. Finally, with the help of hierarchical navigation and reinforcement learning strategies, the robot can intelligently select landing points and dynamically adjust its motion behavior based on local terrain features and mission objectives, guided by a global path, thereby achieving safe, efficient, and fully autonomous exploration in planetary environments with limited energy and communication delays.
[0055] In one embodiment, such as Figure 2 As shown, a control method for a quadrupedal bionic robot applied to planetary exploration, wherein step S1, which involves fusing data from multiple sensors to construct a local elevation map of the terrain and simultaneously estimating the robot's posture, and combining this with preset planetary gravity parameters to construct an environmental cognition model containing information on the environment's geometry and physical characteristics, specifically includes:
[0056] Step S11: Collect three-dimensional point cloud data of the planetary surface using the lidar and stereo vision system on the robot. After processing the three-dimensional point cloud data with filtering and segmentation algorithms, generate a local elevation map of the planetary surface. Mark passable areas and obstacles on the local elevation map to establish a passable area map of the planetary surface.
[0057] Step S12: Read the contact force data from the robot's six-dimensional force sensor on the sole of its foot and the acceleration and angular velocity data from the IMU (Inertial Measurement Unit). Input the read data into an extended Kalman filter for fusion processing to obtain the robot's precise position and three-dimensional attitude in the planetary coordinate system.
[0058] Step S13: Based on the map of traversable areas on the planetary surface and the robot's precise position and three-dimensional posture in the planetary coordinate system, combined with the preset planetary gravity parameters, a comprehensive analysis is performed to construct an environmental cognition model that includes information on environmental geometry (terrain slope, ruggedness) and physical properties (friction coefficient, ground stiffness).
[0059] In step S13, the construction of the environmental cognition model is a process of multi-source information fusion. The system first integrates the traversable area map of the planetary surface generated in step S11 with the robot's precise position and 3D posture provided in step S12, forming a precise perception of the environmental geometry and its own state. Based on this, and combined with preset planetary gravity parameters, a comprehensive analysis algorithm quantifies and extracts key environmental geometric information (such as terrain slope and ruggedness) from the traversable area map, and further calculates the physical characteristics of the environment (such as friction coefficient and ground stiffness). Finally, all this information is integrated into a dynamically updated environmental cognition model, providing crucial decision-making basis for subsequent adaptive gait planning and compliant control.
[0060] In one embodiment, such as Figure 3 As shown, a control method for a quadrupedal bionic robot applied to planetary exploration includes step S2, which involves selecting or adjusting gait parameters online from a gait library based on environmental geometry information, generating the robot's motion trajectory using model predictive control, and simultaneously calculating the foot parabola using a swing leg trajectory generation algorithm to ensure obstacle crossing and impact-free landing. The specific steps include:
[0061] Step S21: Based on the environmental geometric information of the environmental cognition model, select the most suitable basic gait pattern for the current terrain from the gait library, and dynamically adjust the duty cycle and stride parameters of the gait according to the real-time terrain ruggedness and slope data through the fuzzy logic controller to achieve adaptive adjustment of gait parameters.
[0062] Step S22: Using the model predictive control algorithm, based on motion commands (such as going to a specific coordinate point) and the local elevation map, the trajectory of the body's center of mass and the position of the zero moment point in the future time domain are continuously optimized to complete the optimized generation of the body's motion trajectory;
[0063] Step S23: Calculate the foot trajectory for the leg in the swing phase, achieve obstacle crossing through cubic spline curve planning, and use fifth-order polynomial interpolation at the end of the trajectory to ensure that the vertical velocity of the foot returns to zero, generating a safe and smooth foot trajectory to ensure obstacle crossing and impact-free landing.
[0064] The core objective of steps S21 to S23 is to achieve dynamic, stable, and adaptive motion planning for the robot on the complex and unknown planetary surface. Specifically, step S21, based on real-time perception of environmental geometry, intelligently selects and dynamically adjusts gait parameters (such as duty cycle and stride) from a gait library, enabling the robot's basic movement pattern to proactively adapt to changes in terrain slope and ruggedness, laying the foundation for robust movement. Step S22 utilizes a model predictive control (MPC) algorithm to proactively optimize the robot's center of mass trajectory and stability key points (zero-moment points) over a future period based on motion commands and a local map, thereby generating a smooth, stable, and executable body motion trajectory while satisfying dynamic constraints. Step S23, to ensure the safety and effectiveness of each step's landing, plans a foot trajectory for the legs in the swing phase that allows for smooth ground contact in the vertical direction to absorb impact and effective obstacle crossing in the horizontal direction.
[0065] In one embodiment, such as Figure 4 As shown, a control method for a quadrupedal bionic robot applied to planetary exploration, in step S3, involves solving the body pose tracking and foot contact constraints as a quadratic programming problem using a whole-body control framework, adjusting the stiffness and damping of the legs using impedance control to achieve compliant ground contact, and maintaining dynamic balance by managing the phase switching of each leg through a state machine. Specifically, this step includes:
[0066] Step S31: The obtained body motion trajectory, the motion state of each leg and foot end and the ground reaction force are uniformly modeled as a constrained quadratic programming problem, and the optimal torque command of all joints is solved to complete the calculation of the whole body coordinated control quantity.
[0067] Step S32: Based on the principle of impedance control, the stiffness parameters of the legs are dynamically adjusted according to the motion phase of each leg: low stiffness is used in the swing phase to ensure flexibility, high damping mode is switched at the moment of ground contact to absorb impact, and impedance parameters are adaptively adjusted according to the physical characteristic information (such as ground stiffness estimation) in the environmental cognition model in the support phase to achieve a smooth interaction between the foot and the planetary surface.
[0068] Step S33: The switching logic of each leg between the swinging, landing and supporting phases is managed by a finite state machine. The coordination relationship between the four legs is dynamically adjusted according to the real-time attitude error to realize the dynamic balance control of the robot on the planetary surface.
[0069] The core objective of steps S31 to S33 is to accurately translate the upper-level motion trajectory planning into the robot's actual joint movements, and to achieve coordinated control and dynamic balance of the whole body when interacting with complex environments. Specifically, in step S31, through the whole-body control (WBC) framework, the target trajectory of the robot, the contact constraints of each foot, and the overall dynamic model are uniformly solved as a constrained quadratic programming problem, thereby calculating the optimal torque commands for all joints. This ensures that the robot's torso and limbs can operate in a highly coordinated manner as a whole when performing movements. In step S32, using the principle of impedance control, the stiffness and damping parameters of the legs are dynamically adjusted based on the motion phase of the legs (swinging or supporting) and the perceived physical characteristics of the ground (such as stiffness). This allows the robot's feet to act like intelligent springs, compliantly absorbing impacts upon contact with the ground and adapting to terrain changes during support, achieving robust and compliant interaction with planetary surfaces. Step S33 uses a finite state machine to precisely manage the switching logic between the swinging, landing, and supporting phases of each leg, and dynamically adjusts the coordination relationship between the four legs based on the real-time body posture error. This maintains the robot's dynamic balance, enables timely response to external disturbances (such as thrust caused by uneven ground), prevents instability, and ensures continuous and stable movement of the robot on rugged planetary surfaces.
[0070] In one embodiment, such as Figure 5 As shown, a control method for a quadrupedal bionic robot applied to planetary exploration is described. Step S4 employs a hierarchical navigation architecture; global path planning generates an obstacle-avoidance topology route, and local planning updates the landing area in real time based on perception; by evaluating terrain accessibility and energy consumption, a reinforcement learning strategy is used to dynamically adjust the motion behavior. Specifically, this step includes:
[0071] Step S41: Based on the digital elevation map generated from planetary orbit images, the A* algorithm is used to plan the optimal obstacle avoidance topology path connecting multiple exploration points and generate a global navigation route.
[0072] Step S42: At the local navigation level, based on the real-time updated environmental cognition model, assess the safety, slope and slip risk of each potential landing point to complete the selection of a fine landing point;
[0073] Step S43: Invoke the reinforcement learning strategy network that has been trained by a large number of planetary environment simulations, comprehensively analyze the environmental geometry and physical characteristics, motion energy consumption and scientific task priorities (such as the value weight of the detection target) in the environmental cognition model, and make specific behavioral strategies online to achieve energy-optimal intelligent exploration decisions.
[0074] Steps S41 to S43 aim to construct an intelligent navigation decision-making system that balances global efficiency with local safety. The fundamental goal is to enable the robot to move autonomously, reliably, and efficiently on unknown and complex planetary surfaces. Specifically, the system first plans an optimal obstacle-avoidance topology path connecting key exploration points based on a prior map through global path planning, providing macroscopic route guidance for the robot's long journey. Then, at the local level, the system meticulously evaluates the safety of each potential landing point based on real-time perception data, ensuring that the robot's every step is taken in a stable and reliable position, thus translating the macroscopic route into a series of safe, immediate action commands. Finally, by invoking reinforcement learning strategies, the system, like an experienced explorer, comprehensively analyzes terrain geometry and physical characteristics, energy consumption, and the value of the scientific task, making online decisions on the optimal behavior pattern. For example, it might choose to detour around soft sand to save energy, or directly cross high-value rock sample points. This hierarchical navigation architecture works together to ensure that the robot's movement in resource-constrained planetary environments is not only safe but also intelligent and energy-optimized.
[0075] In one embodiment, such as Figure 6 As shown, a control method for a quadrupedal bionic robot applied to planetary exploration also includes:
[0076] Step S5: The robot uses foot force sensors to monitor the interaction force between the foot and the planetary surface in real time. When a sliding tendency is detected, a response strategy is immediately activated. Based on the principle of impedance control, the robot actively enhances the downward pressure of the foot on the planetary surface by increasing the impedance parameter in the vertical direction of the foot, simulating the effect of pushing off the ground to increase the maximum static friction and enhance the ground grip. At the mechanical level, the variable foot structure is controlled according to the environmental geometry and physical characteristics information in the environmental cognition model. When a smooth rock surface is detected, micro-thorns are popped out, and when loose sand is detected, a structure that increases the ground contact area is deployed. Mechanical adaptation is used to provide additional adhesion and improve the robot's traction performance in the low gravity environment of the planet.
[0077] Step S5 aims to address the problem of weak gripping force and slippage caused by insufficient robot weight in low-gravity planetary environments. Step S5 uses foot force sensors to detect slippage trends in real time and immediately activates countermeasures. On the one hand, it actively increases the downward pressure on the feet through impedance control to enhance friction. On the other hand, it controls the variable foot structure (such as popping out thorns on rocks and unfolding structures on sand) to increase adhesion in a mechanically adaptive manner, thereby significantly enhancing the robot's traction performance and ability to pass through complex terrains.
[0078] In one embodiment, such as Figure 7 As shown, a control method for a quadrupedal bionic robot applied to planetary exploration also includes:
[0079] Step S6: Using data from the robot's IMU and joint torque sensors, when the robot is in a stable motion phase, estimate in real time the actual gravitational acceleration and dynamic parameters (such as mass and moment of inertia) of the leg links in the current environment. If the estimated value deviates from the preset planetary gravity parameters, the preset planetary gravity parameters are dynamically corrected, and the robot's mass changes due to carrying or consuming samples are adapted to the changes.
[0080] Based on the identification results, the robot dynamics model relied upon in the whole-body controller (WBC) is dynamically corrected so that it can accurately reflect the changes in robot mass properties caused by differences in planetary environment and sample carrying / consumption, providing a reliable basis for high-precision joint torque solution.
[0081] Step S6 is designed to address the potential errors in the preset planetary gravity parameters and the inaccuracy of the robot's dynamics model due to changes in its own mass caused by carrying samples. Step S6 uses the robot's own sensor data to estimate the actual gravitational acceleration and linkage dynamic parameters in real time and dynamically corrects the model. This design can significantly improve the accuracy of the motion control algorithm and its adaptability to different gravity environments and changes in its own load, ensuring that the robot always operates stably under an accurate robot dynamics model.
[0082] In one embodiment, such as Figure 8 As shown, a control method for a quadrupedal bionic robot applied to planetary exploration also includes:
[0083] Step S7: The robot's posture is continuously monitored by the IMU. When a sudden tilt that exceeds the safety threshold is detected, the preset anti-tipping rescue gait or leg extension action is invoked to prevent the robot from tipping over.
[0084] Step S7 is an active safety measure designed to address the risk that the robot may face sudden loss of posture or even overturning on rugged terrain. Step S7 continuously monitors the robot's posture through the IMU. Once a sudden tilt exceeding the safety threshold is detected, the preset anti-overturning rescue gait or rapid leg extension and retraction movements are immediately invoked to actively restore balance. Its advantage is that it can effectively save the robot from the edge of overturning, greatly enhancing its motion robustness and mission survivability in unknown and dangerous terrain.
[0085] In one embodiment, such as Figure 9 As shown, a control method for a quadrupedal bionic robot applied to planetary exploration also includes:
[0086] Step S8: By analyzing data from the robot's leg joints (such as motor current, temperature, encoder feedback, etc.), it is determined whether the robot's leg joints are overloaded, stuck, or degraded. If a fault is found in a certain leg of the robot (such as a decrease in joint torque), the robot's normal gait is switched to a fault-tolerant gait preset in the gait library (such as a triped limp gait) to maintain the robot's basic mobility and operational capabilities.
[0087] For example, for a completely ineffective leg, the robot can switch to a tripedal limp gait; for a leg with insufficient torque, it can adjust its gait phase and load distribution to reduce its burden.
[0088] The initial design purpose of step S8 is to improve the reliability and durability of the robot in long-term unattended tasks and to cope with failures such as overload, jamming or performance degradation that may occur in key joints. Step S8 diagnoses the health status in real time by analyzing data such as joint motor current and temperature, and automatically switches from the normal gait to the fault-tolerant gait (such as triped limp) when a fault is detected. Its core advantage is that it realizes the control degradation of the system, so that the robot can still maintain basic mobility and operation capabilities after some parts are damaged, thereby significantly extending the task life.
[0089] In one embodiment, such as Figure 10 As shown, a control method for a quadrupedal bionic robot applied to planetary exploration also includes:
[0090] Step S9: Based on the day-night temperature difference on the planetary surface, the robot's thermal management is achieved. When the ambient temperature is higher than the preset first temperature, the maximum output torque of the motor is limited or a low-heat gait (such as "crawling" gait) is selected to prevent overheating. When the ambient temperature is lower than the preset second temperature, the robot's internal temperature is maintained by instructing the joints to perform periodic small movements to prevent the lubricant from solidifying and ensure the normal operation of the joints at low temperatures.
[0091] Step S9 is an intelligent thermal management strategy designed to address the survival and performance threats posed by the extreme diurnal temperature range on the planetary surface to the robot's actuators. At high temperatures, the system is prevented from overheating and being damaged by limiting motor torque or adopting a low-heat gait. At low temperatures, the joints are instructed to perform periodic small movements to maintain internal temperature and prevent lubricant from solidifying. This design ensures that the robot can work safely in a wide temperature range, preventing high-temperature overload to protect the hardware and avoiding low-temperature freezing to ensure the mobility of the joints in severe cold.
[0092] It should be understood that although the steps in the flowcharts of the various embodiments of the present invention are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the various embodiments may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least a portion of the sub-steps or stages of other steps.
[0093] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0094] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.
[0095] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
[0096] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.
Claims
1. A control method for a quadrupedal bionic robot applied to planetary exploration, characterized in that, The control method for this quadrupedal bionic robot applied to planetary exploration includes: By integrating data from multiple sensors to construct a local elevation map of the terrain and simultaneously estimating the robot's posture, and combining it with pre-set planetary gravity parameters, an environmental cognition model containing information on environmental geometry and physical characteristics is constructed. Based on environmental geometry information, gait parameters are selected or adjusted online from the gait database. Model prediction control is used to generate the body's motion trajectory. At the same time, the swing leg trajectory generation algorithm is used to calculate the foot parabola to ensure that obstacles are crossed and landing is impact-free. The whole-body control framework unifies the body posture tracking and foot contact constraints into a quadratic programming problem. Impedance control is used to adjust the stiffness and damping of the legs to achieve compliant ground contact. A state machine is used to manage the phase switching of each leg to maintain dynamic balance. A hierarchical navigation architecture is adopted; global path planning generates obstacle avoidance topology routes, and local planning updates the landing area in real time based on perception; by evaluating terrain accessibility and energy consumption, the movement behavior is dynamically adjusted using reinforcement learning strategies.
2. The control method for a quadrupedal bionic robot applied to planetary exploration according to claim 1, characterized in that, The step of constructing a local terrain elevation map by fusing data from multiple sensors and simultaneously estimating the robot's posture, combined with preset planetary gravity parameters, to build an environmental cognition model containing information on environmental geometry and physical characteristics, specifically includes: The robot uses a lidar and stereo vision system to collect three-dimensional point cloud data of the planet's surface. After the three-dimensional point cloud data is processed by filtering and segmentation algorithms, a local elevation map of the planet's surface is generated. Passable areas and obstacles are marked on the local elevation map to establish a passable area map of the planet's surface. The robot reads the contact force data from the six-dimensional force sensor on the sole of its foot and the acceleration and angular velocity data from the IMU. The read data is then input into an extended Kalman filter for fusion processing to obtain the robot's precise position and three-dimensional attitude in the planetary coordinate system. Based on a map of traversable areas on the planetary surface and the robot's precise position and three-dimensional posture in the planetary coordinate system, combined with pre-set planetary gravity parameters, an environmental cognition model containing information on environmental geometry and physical characteristics is constructed.
3. The control method for a quadrupedal bionic robot applied to planetary exploration according to claim 1, characterized in that, The steps of selecting or adjusting gait parameters online from a gait database based on environmental geometry information, generating the body's motion trajectory using model predictive control, and calculating the foot parabola using a swing leg trajectory generation algorithm to ensure obstacle crossing and impact-free landing specifically include: Based on the environmental geometric information of the environmental cognition model, the most suitable basic gait pattern for the current terrain is selected from the gait library, and the duty cycle and stride parameters of the gait are dynamically adjusted by a fuzzy logic controller according to the real-time terrain ruggedness and slope data, so as to achieve adaptive adjustment of gait parameters. Using a model predictive control algorithm, based on motion commands and local elevation maps, the trajectory of the body's center of mass and the position of the zero moment point in the future time domain are continuously optimized to complete the optimized generation of the body's motion trajectory; The system calculates the foot trajectory for the leg in the swing phase, achieves obstacle crossing through cubic spline curve planning, and uses fifth-order polynomial interpolation at the end of the trajectory to ensure that the vertical velocity of the foot returns to zero, generating a safe and smooth foot trajectory to ensure obstacle crossing and impact-free landing.
4. The control method for a quadrupedal bionic robot applied to planetary exploration according to claim 1, characterized in that, The whole-body control framework unifies the body posture tracking and foot contact constraints into a quadratic programming problem, and uses impedance control to adjust the stiffness and damping of the legs to achieve compliant ground contact. The process of maintaining dynamic balance by managing the phase switching of each leg through a state machine specifically includes: The obtained body motion trajectory, the motion state of each leg and foot end, and the ground reaction force are uniformly modeled as a constrained quadratic programming problem. The optimal torque command for all joints is solved to complete the calculation of the whole-body coordinated control quantity. Based on the principle of impedance control, the stiffness parameters of the legs are dynamically adjusted according to the motion phase of each leg: low stiffness is used in the swing phase to ensure flexibility, high damping mode is switched at the moment of ground contact to absorb impact, and impedance parameters are adaptively adjusted according to the physical characteristics information in the environmental cognition model in the support phase to achieve a smooth interaction between the foot and the planetary surface. By managing the switching logic of each leg between swinging, landing, and supporting phases using a finite state machine, and dynamically adjusting the coordination relationship between the four legs based on real-time attitude errors, the robot achieves dynamic balance control on the planetary surface.
5. The control method for a quadrupedal bionic robot applied to planetary exploration according to claim 1, characterized in that, The system employs a hierarchical navigation architecture; global path planning generates obstacle avoidance topology routes, while local planning updates landing areas in real time based on perception; by evaluating terrain accessibility and energy consumption, it dynamically adjusts movement behavior using reinforcement learning strategies. Specifically, the steps include: Based on digital elevation maps generated from planetary orbit images, the A* algorithm is used to plan the optimal obstacle avoidance topology path connecting multiple exploration points and generate a global navigation route. At the local navigation level, based on the real-time updated environmental cognition model, the safety, slope and slip risk of each potential landing point are assessed to complete the selection of a fine landing point; By invoking a reinforcement learning strategy network trained on extensive planetary environment simulations, and comprehensively analyzing environmental geometry and physical characteristics, motion energy consumption, and scientific task priorities in the environmental cognition model, specific behavioral strategies are determined online, achieving energy-optimal intelligent exploration decisions.
6. The control method for a quadrupedal bionic robot applied to planetary exploration according to any one of claims 1 to 5, characterized in that, Also includes: The robot uses a foot force sensor to monitor the interaction force between its foot and the planetary surface in real time, and immediately initiates a response strategy when a sliding tendency is detected. Based on the principle of impedance control, by increasing the impedance parameter in the vertical direction of the foot, the downward pressure of the foot on the planetary surface is actively enhanced, simulating the effect of pushing off the ground to increase the maximum static friction and enhance the ground grip. At the mechanical level, the variable foot structure is controlled according to the environmental geometry and physical characteristics information in the environmental cognition model. When a smooth rock surface is detected, micro-thorns are popped out, and when loose sand is detected, a structure that increases the ground contact area is deployed. Mechanical adaptation is used to provide additional adhesion and improve the robot's traction performance in the low gravity environment of the planet.
7. The control method for a quadrupedal bionic robot applied to planetary exploration according to claim 1, characterized in that, Also includes: Using data from the robot's IMU and joint torque sensors, the actual gravitational acceleration and dynamic parameters of the leg links in the current environment are estimated in real time when the robot is in a stable motion phase. If the estimated values deviate from the preset planetary gravity parameters, the preset planetary gravity parameters are dynamically corrected, and the robot's mass changes caused by carrying or consuming samples are adapted to the changes.
8. The control method for a quadrupedal bionic robot applied to planetary exploration according to any one of claims 1 to 5, characterized in that, Also includes: The robot's posture is continuously monitored by the IMU. When a sudden tilt that exceeds the safety threshold is detected, the preset anti-tipping rescue gait or leg extension and retraction movements are invoked to prevent the robot from tipping over.
9. The control method for a quadrupedal bionic robot applied to planetary exploration according to claim 8, characterized in that, Also includes: By analyzing data from the robot's leg joints, it is possible to diagnose whether the robot's leg joints are overloaded, stuck, or degraded. If a malfunction is detected in a particular leg, the robot's normal gait is switched to a fault-tolerant gait preset in the gait library to maintain the robot's basic mobility and operational capabilities.
10. The control method for a quadrupedal bionic robot applied to planetary exploration according to any one of claims 1, characterized in that, Also includes: The robot's thermal management is achieved by utilizing the diurnal temperature difference on the planetary surface. When the ambient temperature is higher than a preset first temperature, the maximum output torque of the motor is limited or a low-heat gait is selected to prevent overheating. When the ambient temperature is lower than a preset second temperature, the robot's internal temperature is maintained by instructing the joints to perform periodic small movements to prevent the lubricant from solidifying and ensure the normal operation of the joints at low temperatures.