Sensor-Free Robot Force Sensing Using Jacobian Torque Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing industrial robots require expensive force/torque sensors to sense external forces and torques on their end effectors, which is costly and impractical for simple operations, and existing sensorless methods are computationally intensive or limited by robot manipulator frequency.
Innovation Solution
A method that estimates external forces and torques on an electromechanical actuator-driven robot by identifying gravitational and frictional components in load torques, using a Jacobian matrix to calculate the difference between load torques with and without external forces, and presenting the results in Cartesian space without the need for physical sensors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If force/torque sensors are installed on the robot to measure external forces and torques, then measurement precision is improved, but device complexity and cost increase
Solution Approach 1:
The patent replaces physical force/torque sensors with a computational model-based estimation system. The controller uses the robot's dynamic model, joint torque measurements from encoders, and observed state variables to calculate external forces and torques through mathematical equations, eliminating the need for mechanical sensor installations while achieving comparable measurement precision
Solution Approach 2:
The patent creates a virtual copy of the force/torque sensing function through software-based estimation. By using the robot's dynamic model and available sensor data (joint positions, velocities, accelerations, and motor torques), the system reconstructs the external force/torque information that would otherwise require physical sensors, providing a cost-effective alternative that maintains measurement capability
2Device complexity
If sensorless force estimation methods are used to avoid expensive sensors, then device complexity is reduced, but measurement precision deteriorates due to computational intensity and frequency limitations
Solution Approach 1:
The patent implements a feedback mechanism where the estimated external forces and torques are continuously updated based on actual robot state measurements. The controller uses real-time joint torque data, position, velocity, and acceleration feedback to refine the force estimation, ensuring that measurement precision is maintained through continuous correction rather than being limited by computational frequency
Solution Approach 2:
The patent changes the approach from using complex computational algorithms to solving a system of linear equations with closed-form solutions. By reformulating the force estimation problem to use directly computable parameters from the robot's dynamic model and available sensor data, the system achieves high measurement precision without being constrained by computational intensity or manipulator frequency limitations
3Reliability
If physical force/torque sensors are installed to provide comprehensive plant status information, then reliability is improved, but financial expenses increase
Solution Approach 1:
The patent enables the robot system to self-determine external forces and torques using its own built-in sensors and computational capabilities. The controller utilizes data already available from the robot's motor controllers and encoders, combined with the robot's dynamic model, to generate force feedback information without requiring additional external sensors, thereby maintaining reliability while eliminating sensor costs
Data Source
AI summary
A method for force or torque sensing in an electromechanical actuator-driven robot comprising one or more links, one or more joints, an end effector and a controller is provided, the method comprising: estimating a first set of load torques in one or more joints in a given configuration of the robot without external force or load applied to the end effector; identifying gravitational and frictional components in the first set of load torques; estimating a second set of load torques in the one or more joints in the given configuration of the robot with an external force or load applied to the end effector; calculating a difference between the second set of load torques and the first set of load torques, taking into account the identified gravitational and frictional components; calculating an external force or torque acting on the end effector based on the difference between the second set of load torques and the first set of load torques using a Jacobian matrix for the given configuration of the robot; and presenting the external force or torque in a Cartesian space. An apparatus for force or torque sensing in an electromechanical actuator-driven robot, the apparatus comprising at least one processor programmed to perform said method, a computer program which, when executed by at least one processor, causes the at least one processor to perform force or torque sensing in an electromechanical actuator-driven robot according to said method, and a non-transitory storage medium for storing said program are also provided. The technical result consists in improved precision of force or torque sensing on an end effector of an electromechanical actuator-driven robot in a manner which does not require using expensive force/torque sensors in robot joints.