9results about How to "Guaranteed optimality" patented technology

The invention provides a robot visual positioning and path planning method and system, and the method comprises the steps: extracting static obstacle features and dynamic target motion features from an environment image, and constructing an environment feature map containing time-space correlation information; generating a probability distribution prediction area of the dynamic target; rasterizing and layering a space area between the current pose and the target position of the robot, and endowing grids of different levels with passage cost values; generating an initial path from the current pose to the target position, and performing smooth optimization on the initial path; in the movement process of the robot, monitoring the change of an environment characteristic spectrum, and when it is detected that the actual movement track of the dynamic target deviates from a prediction area and exceeds a deviation threshold value, adjusting a local path section; and calculating a speed and steering control instruction of the robot according to the adjusted local path section, and sending the control instruction to a robot motion control module. According to the invention, the accuracy, safety and real-time performance of robot path planning can be ensured.

This application discloses a model training method, apparatus, device, storage medium, and program product, belonging to the field of machine learning technology. The method generates pseudo-samples based on a first generative model in an adversarial generative network (PGN). The predicted labels of the pseudo-samples are determined by a first discriminative model in the PGN. Based on the predicted and actual labels of the pseudo-samples, the weights of multiple loss functions of the first generative model are iteratively trained until the weights of the multiple loss functions satisfy a first convergence condition. The iterative training of the weights of the multiple loss functions ends, and a first target model is determined based on the weights of the multiple loss functions at the end of the iterative training. This method can learn the most suitable weights for each loss function during iterative training, ensuring the optimality of the weights for each loss function, thereby improving the performance of the final target model.

The invention discloses an optimal path planning method of a pulse coupling neural network based on double constraints. The method comprises the following steps: mapping a path planning environment to a DC-PCNN network; a DC-PCNN neural network model is constructed, and all neurons are initialized; activating a target neuron, and recording the current neuron as a father node; calculating an exponential decay function of the torque deviation, and multiplying the calculated value as a penalty factor by an update item of the internal activity item; calculating a gravitational function value of the flow field constraint, and using the calculated value to update a current neuron dynamic threshold value; comparing the internal activity item with a dynamic threshold; when gt; if yes, activating the neuron, and recording the current neuron as a father node; repeating the steps S4-S6 until the initial neuron is activated; and backtracking all activated nodes, and planning an optimal path. According to the method, the search efficiency is remarkably improved while the path optimality is ensured.

This invention proposes an artificial intelligence-based vehicle control method and system, comprising: acquiring raw data of the vehicle's surrounding environment to construct a digital twin model of the vehicle's surrounding environment; identifying the current driving intention type based on environmental information, historical vehicle driving data, and driver operation behavior characteristics, and calculating a multi-dimensional risk situation index by combining the spatiotemporal correlation between obstacle movement trajectories and the vehicle's current state; performing simulation based on the multi-dimensional risk situation index to generate an adaptive control strategy sequence and convert it into control commands; monitoring the vehicle's actual response parameters after the control commands are executed, and calculating the deviation between the predicted state and the vehicle's actual state; when the deviation exceeds a preset threshold, fine-tuning the deep neural network and reinforcement learning model using the actual response parameters; and storing the optimized model parameters and synchronizing them to other vehicle nodes in the vehicle network. This invention achieves accurate environmental perception and intelligent, flexible generation of control strategies.

This invention belongs to the field of automatic control technology, specifically relating to an autonomous coverage detection and obstacle avoidance control device for a shield tunnel inspection robot. It includes: an inspection sensor assembly, an odometer assembly, a computing and processing assembly, a memory, a motion control assembly, a communication assembly, and a power supply assembly. The inspection sensor assembly includes a vision sensor and a radar sensor. The computing and processing assembly is communicatively connected to the inspection sensor assembly, the odometer assembly, the motion control assembly, the communication assembly, and the power supply assembly. The memory stores instructions that can be executed by the computing and processing assembly. This invention fully utilizes prior knowledge of the shield tunnel structure to improve positioning accuracy, employs a mail carrier coverage algorithm to ensure the integrity of the detection coverage and path optimization, introduces a control obstacle function to achieve real-time obstacle avoidance under safety constraints, and ensures the continuity and reliability of the inspection task in a dynamic environment through the tight coupling of the three components.

The invention discloses an alternating current servo system control method for controlling the stability of an obstacle breaking vehicle body, and the method comprises the steps: constructing an unscented Kalman filter (UKF) load observer based on the UKF, and estimating a load torque value and the rotor angular speed of a permanent magnet synchronous motor in real time through the UKF load observer; feeding back a rotor angular velocity output based on a UKF load observer to an extended state observer (ESO), introducing a load torque value as a feedforward compensation amount into an NLSEF control law, introducing a Newton-Raphson optimization algorithm (NRBO), and performing global optimization on parameters of an active disturbance rejection controller (ADRC) by taking an improved ITAE index as a fitness function; compared with the prior art, the method can improve the tracking precision and the anti-interference performance of the barrier-breaking vehicle under the complex dynamic working condition, and guarantees the stability of the barrier-breaking vehicle body in the advancing process.

This application discloses a method, system, device, equipment, and medium for controlling the facial expressions of a bionic robot, relating to the field of simulated robot control. The method includes: constructing a virtual model of the bionic robot, setting several virtual attachment points, acquiring a continuous facial expression sequence, extracting several target facial expression moments and corresponding target facial expression states from the continuous facial expression sequence; setting the target facial expression state at each target facial expression moment as the target to be imitated, using the displacement parameters of each servo motor as optimization variables, and constructing an objective function based on the geometric error between the actual position of the virtual attachment point in the virtual model and the target position under the target facial expression state; iteratively sampling and simulating in the parameter space that satisfies the servo motor constraints, calculating the objective function values ​​corresponding to the displacement parameters, until the optimal servo motor parameter set at the corresponding target facial expression moment is obtained, and combining it into a continuous control sequence; converting the continuous control sequence into facial expression control commands, and deploying them to the servo motors of the bionic robot for execution. This achieves optimization of the bionic head's facial expressions.

The invention provides a natural gas P2 heavy load vehicle-oriented EMS adaptive mixed domain dynamic planning method, and relates to the field of new energy. The invention provides a self-adaptive hybrid domain dynamic planning method. Through systematic layering, the state and action space is greatly reduced on the premise of maintaining the calculation precision. In order to solve the problem that a large-capacity battery is neglected in discretization calculation due to small single-step SOC change, a coarse and fine combined adaptive SOC grid is constructed to dynamically adjust a calculation grid. Three-dimensional fitting is carried out on multiple gears of a motor and an engine based on a physical model so as to establish an optimal gear shifting model, and then the current optimal gear is rapidly locked according to the accelerator opening degree and the vehicle speed. And the optimal motor-engine torque combination is quickly screened in a mode of combining gradient descent and global search. By adopting the AHD-DP method disclosed by the invention, excellent robustness, optimality and generalization ability are shown.

The application discloses a robot autonomous navigation method and device fusing a patrol strategy, comprising the following steps: matching and screening from a pre-constructed feature vector library through a preset patrol rule, and extracting key nodes; performing cluster analysis on the extracted key nodes, and removing redundant nodes with too close distances to obtain topological nodes; analyzing the spatial connectivity among the topological nodes, establishing edge association among the nodes, and generating a topological graph; calculating the shortest comprehensive cost path from a start matching node to an end matching node in the topological graph to obtain a global node sequence; performing interpolation processing on the global node sequence, and generating a continuous and smooth trajectory curve as a navigation path of the robot. The application realizes environment space dimension reduction through construction of a topological graph, and fuses a preset patrol rule to perform node extraction and path planning, so that the search efficiency is significantly improved, and the calculation amount is reduced while the optimal path is ensured.