Practical formulation of obstacle avoidance in the Task-Priority framework for use in robotic inspection and intervention scenarios

Abstract

This paper presents a new formulation of a reactive obstacle avoidance algorithm, in the Task-Priority framework, delivering a practical solution for obstacle avoidance between vehicle-manipulator systems and complex environments. The presented concepts were implemented on an intervention autonomous underwater vehicle (I-AUV) and tested in an underwater pipe structure inspection and valve turning scenario, in a test tank, using GIRONA500 with an ECA 5E Micro manipulator. The obstacle avoidance is treated as an inequality (set-based) task, which takes into account all obstacles that are interacting with the robot links. Both the robot and the obstacles are represented by spheres to allow for analytical formulation. However, the environment is wrapped with spheres based on its actual geometry stored as an Octomap, hence it can be represented at different resolutions. Depending on the type of the mission performed by the robot we defined two modes of operation: (1) Navigation and Inspection, and (2) Intervention. For each mode, the algorithm takes into account different number of key points at the I-AUV and a different resolution of the environment representation. Typically, for the Intervention mode the resolution is higher, to allow for more precise motion. We also present an escape point strategy in case of the robot getting stuck between obstacles.