Collision Detection and Isolation on a Robot using Joint Torque Sensing

Abstract

As robotic systems become more flexible and intelligent, they must be able to move into environments with a high degree of uncertainty or clutter, such as our homes, workplaces, and the outdoors. In these unstructured scenarios, it is possible that the body of the robot collides with its surroundings. As such, it would be desirable to characterise these contacts in terms of their location and interaction forces. This paper addresses the problem of detecting and isolating collisions between a robotic manipulator and its environment, using only on-board joint torque and position sensing. The algorithm is based on a particle filter and, under some assumptions, is able to identify the contact location anywhere on the robot body. It requires the robot to perform small exploratory movements, progressively integrating the new sensing information through a Bayesian framework. The approach was tested and benchmarked in simulation, with respect to its accuracy and robustness. Validation using a robot with joint torque sensing in a real environment demonstrated the applicability of the method to real-world scenarios.

Collision Detection and Contact Point Estimation Using Virtual Joint Torque Sensing Applied to a Cobot

“…When comparing our work to the results presented in the literature, we can state that we achieve a comparable accuracy. In [10], a CPF is applied on a KUKA LBR iiwa experimentally, where a contact point accuracy below 2 cm is achieved on link 6, using 10000 particles.…”

“…The authors employ the CPF and the residual method on an Atlas humanoid robot in simulation and are able to estimate up to 3 simultaneous contact points accurately. In [10], experimental results of the application of the CPF without the residual method on a KUKA LBR iiwa 7-dof robot are shown, relying only on joint torque sensors and position measurements.…”

Autonomous Unwrapping of General Pallets: A Novel Robot for Logistics Exploiting Contact-Based Planning

“…The different contact isolation and identification methods presented in the literature may be classified based on (i) the nature of the algorithm, the most common classes being model-based analytical methods [30]- [32] and learning methods [33]- [35]; (ii) the sensors used for the estimation, usually either proprioceptive sensors, such as joint position and velocity sensors [32] or joint torque sensors [33], [34], [36], or external force and torque sensors [31], [37], vision sensors [38], and tactile sensors [35]. In general, relying on proprioceptive sensors requires a less complex and less expensive setup.…”

Multi-Scenario Contacts Handling for Collaborative Robots Applications

“…After this publication, several other research groups investigated similar methods. In [15] the motion model of particles was modified with a combination of small self-movements of the robot to increase accuracy. Monte Carlo localization was used in [16] and compared to direct optimization and machine learning methods.…”