TY - GEN
T1 - Sampling based motion planning with reachable volumes
T2 - 2014 IEEE/RSJ International Conference on Intelligent Robots and Systems, IROS 2014
AU - McMahon, Troy
AU - Thomas, Shawna
AU - Amato, Nancy M.
N1 - Publisher Copyright:
© 2014 IEEE.
PY - 2014/10/31
Y1 - 2014/10/31
N2 - Reachable volumes are a geometric representation of the regions the joints of a robot can reach. They can be used to generate constraint satisfying samples for problems including complicated linkage robots (e.g. closed chains and graspers). They can also be used to assist robot operators and to help in robot design.We show that reachable volumes have an O(1) complexity in unconstrained problems as well as in many constrained problems. We also show that reachable volumes can be computed in linear time and that reachable volume samples can be generated in linear time in problems without constraints. We experimentally validate reachable volume sampling, both with and without constraints on end effectors and/or internal joints. We show that reachable volume samples are less likely to be invalid due to self-collisions, making reachable volume sampling significantly more efficient for higher dimensional problems. We also show that these samples are easier to connect than others, resulting in better connected roadmaps. We demonstrate that our method can be applied to 262-dof, multi-loop, and tree-like linkages including combinations of planar, prismatic and spherical joints. In contrast, existing methods either cannot be used for these problems or do not produce good quality solutions.
AB - Reachable volumes are a geometric representation of the regions the joints of a robot can reach. They can be used to generate constraint satisfying samples for problems including complicated linkage robots (e.g. closed chains and graspers). They can also be used to assist robot operators and to help in robot design.We show that reachable volumes have an O(1) complexity in unconstrained problems as well as in many constrained problems. We also show that reachable volumes can be computed in linear time and that reachable volume samples can be generated in linear time in problems without constraints. We experimentally validate reachable volume sampling, both with and without constraints on end effectors and/or internal joints. We show that reachable volume samples are less likely to be invalid due to self-collisions, making reachable volume sampling significantly more efficient for higher dimensional problems. We also show that these samples are easier to connect than others, resulting in better connected roadmaps. We demonstrate that our method can be applied to 262-dof, multi-loop, and tree-like linkages including combinations of planar, prismatic and spherical joints. In contrast, existing methods either cannot be used for these problems or do not produce good quality solutions.
UR - http://www.scopus.com/inward/record.url?scp=84911466473&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84911466473&partnerID=8YFLogxK
U2 - 10.1109/IROS.2014.6943082
DO - 10.1109/IROS.2014.6943082
M3 - Conference contribution
AN - SCOPUS:84911466473
T3 - IEEE International Conference on Intelligent Robots and Systems
SP - 3705
EP - 3712
BT - IROS 2014 Conference Digest - IEEE/RSJ International Conference on Intelligent Robots and Systems
PB - Institute of Electrical and Electronics Engineers Inc.
Y2 - 14 September 2014 through 18 September 2014
ER -