TY  - JOUR
PY  - 2021//
TI  - Repulsion-oriented reciprocal collision avoidance for multiple mobile robots
JO  - Journal of intelligent and robotic systems
A1  - Wang, Shaojie
A1  - Hu, Xiaoguang
A1  - Xiao, Jin
A1  - Chen, Tianyou
SP  - e12
EP  - e12
VL  - 104
IS  - 1
N2  - In this paper, a new collision avoidance algorithm based on velocity obstacles (VO) is proposed for distributed mobile robots to achieve oscillation-free autonomous navigation, which is called repulsion-oriented reciprocal collision avoidance (RORCA). Firstly, the problem of reciprocal collision avoidance for autonomous robots is defined. Each robot makes independent decisions by sensing the position and velocity of its surrounding objects without centralized coordination or communication with other robots. The final aims of the robots are to reach some pre-defined goals. Secondly, the VO algorithm and its variants are reviewed. Their potential problems in trajectory smoothness and navigation safety are used as test elements of the new algorithm. The RORCA algorithm considers the virtual repulsion direction of the paired robots and the reciprocity of collision avoidance between them to constructs a feasible velocity set for each robot. This set is proved to be a non-empty convex set containing at least zero vector. In this way, the local motion planning of the robot is transformed into a convex optimization problem for solving the optimal velocity on this set. For ease of explanation, we elaborate the algorithm in a two-dimensional (2D) workspace, which takes into account the kinematics constraints and measurement uncertainty from sensors. Then, the expansion of the RORCA in three-dimensional (3D) space is introduced. Finally, comparative simulations are performed in 2D and 3D scenarios to demonstrate the effectiveness of the proposed technique. The results show that the RORCA algorithm is particularly suitable for solving the problems trajectory oscillation, reciprocal dance and potential collisions of robots in dense environments, and it has excellent performance in safety and computational efficiency.<p />  <p>Language: en</p>
LA  - en
SN  - 0921-0296
UR  - http://dx.doi.org/10.1007/s10846-021-01528-6
ID  - ref1
ER  -