Multi-arm manipulation has the potential to dramatically improve the throughput, efficiency, and dexterity of many manipulation tasks. Recent work coordinates multiple robot arms in a shared workcell by allocating and scheduling tasks for greater parallelism, jointly planning multi-arm motions, or directly learning multi-arm policies. So far, these approaches are either too slow, rely on domain-specific assumptions, or scale poorly with the number of robots and task horizon. Thus, we introduce cuArms, a conflict-based planner co-designed with GPU parallelism that efficiently decomposes the search space using reusable multi-modal roadmaps for individual robots and resolves conflicts and deadlocks through a lazy search. To jointly optimize task assignment, schedules, motions, and manipulation skills, we integrate a high-level scheduler that proposes candidate skeletons for cuArms to ground into coordinated manipulation plans. Experimental results show that our method establishes a new state of the art, with real-time initial solutions and anytime improvements in both motion quality and scheduling, without days of domain-specific training required by learning approaches. Videos of both simulated and real-world industrial tasks are available on https://cuarms.github.io/.
Figures & tables
Fig. 1 : Applications of cuArms : a) tabletop manipulation, b) welding, c) real object reorientation, and d) bimanual carry.
Fig. 2 : System and cuArms overview: A human or agent defines the task: the robots and their kinematics, the objects, goals, and constraints. The domain also supplies samplers that can generate feasible manipulation choices. A task scheduler, such as ScheduleStream, proposes a candidate schedule. This includes task assignment and ordering, along with satisfying samples such as placements, grasps, and inverse kinematic solutions. cuArms builds a per-robot multi-modal roadmap, searches for conflict-free robot movements and skills with GPU-batched conflict tables, and refines the solution trajectory with shortcutting. cuArms can iteratively replan with different schedule skeletons to improve upon infeasible or poor skeletons.
Fig. 3 : Example schedule skeletons in the four domains from Fig. 1 : (a) tabletop pick-and-place, (b) multi-robot reaching for welding, (c) bimanual handover, and (d) bimanual pick and coordinated carry. The high-level scheduler produces the skeleton with corresponding skill parameters such as handover poses, object placements, and grasp poses. cuArms then computes the coordinated multi-arm motions and skills based on the schedule skeleton.
Figure 4
Fig. 5 : RoboBallet schedule search and refinement. Execution makespan improves as cuArms shortcuts initial trajectories and searches alternative schedules.
Targets
Robots
Method
Solved
toffline(s)
t0 (s)
J0 (s)
Jf (s)
10
4
cuArms
150/150
0.81
0.20
10.61
5.64
40
4
cuArms
150/150
0.81
0.41
31.16
16.56
40
8
cuArms
150/150
1.74
0.87
25.35
14.87
10
4
RL
29/150
1.6×105
0.11
–
6.00
40
4
RL
50/150
2.2×105
0.42
–
24.25
TABLE II : RoboBallet scheduling and planning. toffline is the median base roadmap construction time for cuArms and training time for RL. t0 is the total inference time for RL.
4 arms
8 arms
Targets
10
40
40
Greedy Scheduler
0.59
5.12
3.06
Roadmap Construction
98.70
155.39
271.17
Heuristics
47.68
120.81
146.44
Candidate Expansion & Sort
8.24
24.24
32.06
Conflict Table
22.46
65.94
277.46
TABLE III : cuArms median time-to-initial solution (ms) breakdown in RoboBallet. Roadmap construction refers to connecting new weld targets to the base roadmap. Collision checking includes queries in roadmap construction and the conflict table.
Fig. 6 : Tabletop manipulation refinement. cuArms improves plans across skeletons and manipulation modes. Makespan can increase while the weighted optimization objective improves.
cuArms
ScheduleStream
Task
Solved
t0
J0
Jf
Solved
t0
J0
Hold
45/45
0.48
5.46
2.37
44/45
1.75
3.74
Stack
45/45
2.26
12.84
6.67
38/45
4.02
16.11
Pack
45/45
2.96
12.02
6.96
38/45
4.54
17.45
Unpack
45/45
1.60
11.50
6.55
39/45
3.30
13.25
TABLE IV : Multi-arm TAMP for tabletop manipulation with two to four objects. J0 and Jf are the initial and final makespan, respectively.
Fig. 7 : Scaling with number of objects. cuArms finds more efficient manipulation plans in less time than ScheduleStream, which also fails to find any solutions for 8-object packing.
Fig. 8 : Bimanual Electronic Assembly Tasks with cuArms .
Multi-robot-arm motion planning is a key challenge in deploying multiple manipulators for industrial tasks such as manufacturing. Existing search-based and sampling-based solvers often require significant computation time to produce collision-free, high-quality motions suitable for safe real-world execution. In this work, we introduce a new suite of multi-robot-arm motion planners capable of near real-time motion generation, combining classical planning algorithms with state-of-the-art vectorized collision-checking techniques. Based on CPU SIMD instructions, our new planners accelerate their primary bottleneck, collision checking, and achieve up to two orders of magnitude speedup in both motion planning and execution postprocessing for multi-arm manipulation tasks. We also release our implementation to lower the barrier for research and development of multi-robot-arm planning and manipulation problems. Code is available at https://vamp-mr.github.io/vamp-mr
Philip Huang, Chenrui Gao, Jiaoyang Li
Robotics Institute, Carnegie Mellon University, Pittsburgh, PA 15213, USA · University of Michigan, Ann Arbor
This paper presents Parallel ARC (P-ARC), a parallel variant of the Adaptive Robot Coordination (ARC) approach to multi-robot motion planning (MRMP). P-ARC proposes a parallel variant for each of the three main stages in ARC: initial individual solutions, conflict detection, and conflict resolution, exploiting the independence created by ARC's decomposition of the MRMP problem. Additionally, we employ an OR-parallel multi-start strategy to both ARC and P-ARC, creating a hybrid parallel strategy OR-P-ARC. We evaluate the impact of the different parallel strategies for ARC using a set of scaling 2D mobile and planar manipulator scenarios with up to 128 robots to control for conflicts and work distribution across the stages of ARC. Additionally, we demonstrate planning time speedups approaching 4X over the sequential version for large Panda multi-manipulator teams in real-world inspired scenarios when deploying 16 CPU cores.
James D. Motes, Marco Morales, Nancy M. Amato
Parasol Lab, School of Computing and Data Science, University of Illinois at Urbana Champaign, Champaign, IL, 61820 USA · Department of Computer Science at Instituto Tecnológico Autónomo de México (ITAM), Mexico City, México
Learned manipulation policies increasingly predict motions for abstract "hands" and are attractive in practice because they rely on easily collected demonstrations and transfer across robot platforms. Executing these trajectories on multi-arm robots, however, is not trivial. Multi-hand policy outputs must be assigned to physical arms, each arm must realize a configuration-space motion that tracks its prescribed end-effector trajectory, and all arms must respect kinematic limits and avoid collisions. In the absence of algorithms that directly address this problem, practitioners typically extend single-arm inverse-kinematics (IK) pipelines in an ad hoc way, with no guarantees of feasibility or safety. In this work, we close this execution gap with a search-based framework that is theoretically complete for grounding policy-generated multi-hand trajectories onto physical multi-arm systems. Building on Conflict-Based Search, our method explicitly searches over both the discrete assignment of trajectories to arms and the continuous Jacobian null spaces of redundant manipulators, using redundancy to avoid inter-arm collisions while tracking the prescribed motions. This unified treatment of assignment and null-space motion yields a practically efficient planner that safely realizes coordinated manipulation-policy outputs on multi-arm robots. See omcbsa.github.io for more.
Yorai Shaoul, Jiaoyang Li, Maxim Likhachev
The Robotics Institute, Carnegie Mellon University