Fast Coordinated Bimanual Motion Planning With Hard Constraints

Borna Paro; Luka Petrović; Ivan Marković

cs.ROAug 21, 2026

Fast Coordinated Bimanual Motion Planning With Hard Constraints

Authors: Borna Paro, Luka Petrović, Ivan Marković

Organizations: University of Zagreb Faculty of Electrical Engineering and Computing, Laboratory for Autonomous Systems and Mobile Robotics (LAMOR), Unska 3, HR-10000, Zagreb, Croatia

Abstract

Bimanual manipulation enables complex tasks but introduces added complexity from the high number of degrees of freedom involved. When handling rigid objects, the relative transformation between the two end effectors must remain fixed throughout the motion, manifesting as a nonlinear equality constraint that confines the feasible configuration space to a measure-zero manifold and challenges conventional motion planners. We propose a fast bimanual motion planning pipeline that enforces this hard transformation constraint continuously along the entire path, using a leader-follower parameterization: the leader's configuration is treated as a free variable, while the follower's is determined via inverse kinematics to satisfy the constraint. We extensively evaluate the method in simulation across diverse environments, constraints and bimanual platforms, achieving 19.4x faster planning than prior work while guaranteeing continuous constraint satisfaction. Real-world experiments on a bimanual Kinova Gen3 setup, involving tray transport and elongated-object manipulation, validate direct transfer of planned trajectories to physical hardware.

Figures & tables

Panel of Figure 1 — Fig. 1 : Real-world demonstrations of constrained coordinated bimanual motion planning. The two robot arms jointly transport (a) a tray while maintaining it upright to prevent spilling the water and tipping the object, and (b) an elongated object while preserving the coordinated dual-arm grasp.

Fig. 2 : Overview of the proposed constrained bimanual planning pipeline.

Panel of Figure 3 — Fig. 3 : Trajectories for three different start-goal configurations using the KUKA iiwa bimanual setup.

Method	Metric	T $\to$ M	M $\to$ B	B $\to$ T
IK-PRM [ 5 ]	Length	4.67	8.93	9.21
	Time (s)	0.46	0.64	0.61
IK-GCS [ 5 ]	Length	2.09	3.32	5.62
	Time (s)	3.41	2.32	3.32
IK-BiRRT [ 5 ]	Length	9.91	8.69	11.42
	Time (s)	49.42	52.53	54.10

TABLE I : Path length (in C-space) and online planning time (in seconds) on KUKA iiwa benchmark. Methods above the dashed line require offline computation; methods below are fully online. Bold marks the best value in each column; underline marks the best among the fully-online methods.

Constr.	Planner	PSR	$t_{\text{plan}}$	$\ell$	$t_{\text{pipe}}$	PSR	$t_{\text{plan}}$	$\ell$	$t_{\text{pipe}}$	PSR	$t_{\text{plan}}$	$\ell$	$t_{\text{pipe}}$
		No obst.				Sparse				Dense
Kinova Gen3
$\mathcal{T}_{1}$	Ours	98.0	$0.98$	$10.3$	$3.75$	80.0	$22.02$	$26.9$	$30.01$	86.7	$14.58$	$29.1$	$24.98$
$\mathcal{T}_{1}$	Baseline	$32.0$	$1.78$	$16.8$	$3.57$	$0.0$ †	$3.07$	$17.8$	—	$0.0$ †	$2.97$	$22.8$	—
$\mathcal{T}_{2}$	Ours	93.0	$1.06$	$9.5$	$2.92$	70.0	$29.94$	$31.2$	$27.85$	93.3	$17.63$	$15.5$	$21.10$
$\mathcal{T}_{2}$	Baseline	$49.0$	$2.47$	$14.1$	$3.18$	$6.7$	$3.62$	$20.9$	$4.97$	$23.3$	$3.54$	$18.4$	$3.55$

TABLE II : Full quantitative results on both platforms. PSR : pipeline success rate (%).

t_{\text{plan}}

: average path-planning time (s) over successful paths,

\ell

: average path length in the full configuration space.

t_{\text{pipe}}

: average end-to-end pipeline time (s) for successful trajectories. Bold PSR marks our method.

\dagger

no valid trajectory for any query; “—” no successful pipeline.

Fig. 4 : Real-world tray transport experiment. In addition to the relative transform constraint, global orientation constraint needed to be satisfied. The motion is best viewed from supplementary videos.

Fig. 5 : Real-world elongated object transport experiment. Movement sequence: blue $\rightarrow$ green $\rightarrow$ red. The motion is best viewed from supplementary videos.

Motion Planning

Motion planning finds safe, feasible paths and trajectories for robots and vehicles amid geometric, dynamic, and task constraints.

See all 410 papers

Vectorizing Projection in Manifold-Constrained Motion Planning for Real-Time Whole-Body Control
Apr 14, 2026Shrutheesh R Iyer, I-Chia Chang, Andrew Z. Liu +2Motion Planning Constrained Motion Planning
Department of Computer Science, Purdue University · School of Mechanical Engineering, Purdue University
CAMP: Cooperative Arm-Hand Motion Planning in Constrained Spaces
Sep 24, 2026Ziyuan Wang, Yunlong Shan, Fei Mo +6Constrained Motion Planning Robot Motion Generation
Department of Mechanical Engineering and Automation, Harbin Institute of Technology, Shenzhen, China · School of Advanced Engineering, Great Bay University, Dongguan, Guangdong, China · State Key Laboratory for Multi-target Natural Medicine, China Pharmaceutical University, Nanjing, China +4
Unified Learning of Temporal Task Structure and Action Timing for Bimanual Robot Manipulation
Mar 6, 2026Christian Dreher, Patrick Dormanns, Andre Meixner +1Bimanual Robotic Manipulation Temporal Reasoning
Institute for Anthropomatics and Robotics, Karlsruhe Institute of Technology, Karlsruhe, Germany

Apr 14, 2026J/K move · Enter open · S save

Vectorizing Projection in Manifold-Constrained Motion Planning for Real-Time Whole-Body Control

Shrutheesh R Iyer, I-Chia Chang, Andrew Z. Liu, Yan Gu+1

Department of Computer Science, Purdue University · School of Mechanical Engineering, Purdue University

Motion Planning Constrained Motion Planning Whole-Body Control

Open paper AskPDF arXiv

Many robot planning tasks require satisfaction of one or more constraints throughout the entire trajectory. For geometric constraints, manifold-constrained motion planning algorithms are capable of planning collision-free path between start and goal configurations on the constraint submanifolds specified by task. Current state-of-the-art methods can take tens of seconds to solve these tasks for complex systems such as humanoid robots, making real-world use impractical, especially in dynamic settings. Inspired by recent advances in hardware accelerated motion planning, we present a CPU SIMD-accelerated manifold-constrained motion planner that revisits projection-based constraint satisfaction through the lens of parallelization. By transforming relevant components into parallelizable structures, we use SIMD parallelism to plan constraint satisfying solutions. Our approach achieves up to 100-1000x speed-ups over the state-of-the-art, making real-time constrained motion planning feasible for the first time. We demonstrate our planner on a real humanoid robot and show real-time whole-body quasi-static plan generation. Our work is available at https://commalab.org/papers/mcvamp/.

Abstract

Motion Planning

Explore similar work

Vectorizing Projection in Manifold-Constrained Motion Planning for Real-Time Whole-Body Control

CAMP: Cooperative Arm-Hand Motion Planning in Constrained Spaces

Unified Learning of Temporal Task Structure and Action Timing for Bimanual Robot Manipulation