# Differentiable GPU-Parallelized Task and Motion Planning

arXiv: https://arxiv.org/abs/2411.11833 · HTML: https://mlsanity.com/paper/2411.11833/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 18 Nov 2024.

## Built on by (most cited first, 7 of 7)

- [A Systematic Study of Large Language Models for Task and Motion Planning With PDDLStream](https://mlsanity.com/paper/2510.00182/) (2025)
- [DREAM: Deployment-Time Demonstration Generation via Real-to-Sim for Scalable Policy Adaptation](https://mlsanity.com/paper/2608.29078/) (2026)
- [RoboPlayground: Democratizing Robotic Evaluation through Structured Physical Domains](https://mlsanity.com/paper/2604.05226/) (2026)
- [Grounded World Model for Semantically Generalizable Planning](https://mlsanity.com/paper/2604.11751/) (2026)
- [Rational Inverse Reasoning: Few-Shot Imitation by Inferring Intent through Planning](https://mlsanity.com/paper/2508.08983/) (2025)
- [Learning to Plan & Schedule with Reinforcement-Learned Bimanual Robot Skills](https://mlsanity.com/paper/2510.25634/) (2025)
- [ScheduleStream: Temporal Planning with Samplers for GPU-Accelerated Multi-Arm Task and Motion Planning & Scheduling](https://mlsanity.com/paper/2511.04758/) (2025)

## Abstract (arXiv)

Planning long-horizon robot manipulation requires making discrete decisions about which objects to interact with and continuous decisions about how to interact with them. A robot planner must select grasps, placements, and motions that are feasible and safe. This class of problems falls under Task and Motion Planning (TAMP) and poses significant computational challenges in terms of algorithm runtime and solution quality, particularly when the solution space is highly constrained. To address these challenges, we propose a new bilevel TAMP algorithm that leverages GPU parallelism to efficiently explore thousands of candidate continuous solutions simultaneously. Our approach uses GPU parallelism to sample an initial batch of solution seeds for a plan skeleton and to apply differentiable optimization on this batch to satisfy plan constraints and minimize solution cost with respect to soft objectives. We demonstrate that our algorithm can effectively solve highly constrained problems with non-convex constraints in just seconds, substantially outperforming serial TAMP approaches, and validate our approach on multiple real-world robots. Project website and code: https://cutamp.github.io