US 12,103,187 B2
Path planning method and biped robot using the same
Xingxing Ma, Shenzhen (CN); Chunyu Chen, Shenzhen (CN); Ligang Ge, Shenzhen (CN); Yizhang Liu, Shenzhen (CN); Hongge Wang, Shenzhen (CN); Jie Bai, Shenzhen (CN); Zheng Xie, Shenzhen (CN); Jiangchen Zhou, Shenzhen (CN); Meihui Zhang, Shenzhen (CN); Shuo Zhang, Shenzhen (CN); and Youjun Xiong, Shenzhen (CN)
Assigned to UBTECH ROBOTICS CORP LTD, Shenzhen (CN)
Filed by UBTECH ROBOTICS CORP LTD, Shenzhen (CN)
Filed on Nov. 2, 2021, as Appl. No. 17/516,729.
Claims priority of application No. 202011598911.0 (CN), filed on Dec. 29, 2020.
Prior Publication US 2022/0203534 A1, Jun. 30, 2022
Int. Cl. B25J 9/16 (2006.01); B62D 57/032 (2006.01); G05D 1/43 (2024.01); G05D 1/622 (2024.01); G05D 1/644 (2024.01); G05D 109/12 (2024.01)
CPC B25J 9/1666 (2013.01) [B62D 57/032 (2013.01); G05D 1/43 (2024.01); G05D 1/637 (2024.01); G05D 1/644 (2024.01); G05D 2109/12 (2024.01)] 18 Claims
OG exemplary drawing
 
1. A computer-implemented path planning method for a biped robot, comprising:
generating a candidate node set for a next foot placement based on one or more robot parameters of the biped robot and joint information of a current node;
adding one or more valid candidate nodes, not colliding with an obstacle, from the candidate node set to a priority queue based on obstacle information, and calculating a cost value of each of the valid candidate nodes in the priority queue and taking the valid candidate node with the smallest cost value as an optimal node to output;
determining whether the outputted optimal node and a target node meet a preset distance condition;
in response to the outputted optimal node and the target node not meeting the preset distance condition, returning to the step of generating the candidate node set for the next foot placement, to generate a candidate node set for a next foot placement of the outputted optimal node, until the outputted optimal node and the target node meet the preset distance condition; and
in response to the outputted optimal node and the target node meeting the preset distance condition, generating a sequence of optimal output nodes to approach the target node according to all the outputted optimal nodes, and controlling the biped robot to move to the target node from an initial node according to the sequence of optimal output nodes to approach the target node;
wherein the priority queue is a data structure comprising cost values of the candidate nodes in the priority queue calculated using a cost evaluation function, and the cost values are sorted from small to large; and
the method further comprises:
generating a global path from the initial node to the target node according to navigation information and a preset expansion radius of the biped robot using a preset path planning algorithm;
wherein the global path is for at least one of: optimizing the cost evaluation function, and reducing a number of the candidate nodes in the candidate node set;
wherein the cost evaluation function comprises: a heuristic value and a spent cost of each of the candidate nodes in the priority queue, the heuristic value refers to a predicted cost from a current candidate node in the priority queue to the target node, and the spent cost value refers to a spent cost from the initial node to the current candidate node;
wherein the cost evaluation function is optimized using the global path by calculating the heuristic value of each of the candidate nodes in the priority queue using the global path; and
wherein the heuristic value of the current candidate node is calculated by:
traversing all nodes on the global path from the initial node to the target node; and
in response to a line between a candidate node on the global path and the current candidate node being perpendicular to a tangent direction of the global path at the candidate node, calculating a length of the line, calculating a distance from the candidate node on the global path to the target node on the global path, calculating a sum of the length of the line and the distance, and taking the sum as the heuristic value of the current candidate node.