Learning Geometric Reasoning Networks for Robot Task and Motion Planning

We present Geometric Reasoning Networks (GRN), a graph neural network-based model for predicting action and grasp feasibility in Task and Motion Planning (TAMP). GRN reduces reliance on geometric planners by incorporating interpretability mechanisms such as inverse kinematics feasibility prediction and grasp obstruction estimation, enabling efficient and explainable planning in complex 3D environments.

The Thirteenth International Conference on Learning Representations (ICLR 2025)
Extending Task and Motion Planning with Feasibility Prediction: Towards Multi-Robot Manipulation Planning of Realistic Objects

We introduce a multi-robot TAMP algorithm that leverages feasibility prediction to address complex manipulation tasks. Our approach extends previous methods by enabling the handling of mesh-shaped objects and collaborative multi-robot settings, demonstrating significant improvements over baseline methods.

IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2024)
Learning to Predict Action Feasibility for Task and Motion Planning in 3D Environments

We introduce a novel approach for Task and Motion Planning (TAMP) in 3D environments, combining an efficient 3D scene representation with a deep neural network to predict action feasibility, significantly reducing geometric planning time by up to 90% on complex problems.

2023 IEEE International Conference on Robotics and Automation (ICRA)

Simultaneous Action and Grasp Feasibility Prediction for Task and Motion Planning through Multi-Task Learning

LAAS-CNRS
2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)

Given a task and motion planning (TAMP) problem, our method leverages AGFP-Net, a multi-task neural network, to predict both action feasibility and the feasibility of various grasp types. These predictions are used as heuristics to guide the symbolic planner towards geometrically feasible solutions, significantly reducing the time spent on geometric planning. This approach enables the robot to solve complex manipulation tasks more efficiently, even when goals are not fully specified.

Abstract

In this paper, we address task and motion plan-ning (TAMP) which is an important yet challenging robotics problem. It is known to suffer from the high combinatorial complexity of discrete search, often requiring a large number of geometric planning calls. We build upon recent works in TAMP by taking advantage of learning methods to provide action feasibility information as a heuristic to the symbolic planner, thus guiding it to a geometrically feasible solution and reducing geometric planning time. We propose AGFP-Net, a multi-task neural network predicting not only action feasibility, but also the feasibility of a set of grasp types. We also propose an improved feasibility-informed TAMP algorithm capable of solving more complex problems, and handling goals which are not fully specified. Comparative results obtained on different problems of varying complexity show that our method is able to greatly reduce task and motion planning time.

Proposed Model

Diagram of the proposed neural network AGFPNet

A visualization of the proposed neural network architecture used for action and grasp feasibility prediction.

Benchmarks

Reorder Benchmark

Reorder problem: Objects initially placed on a two-shelf cupboard must be moved to another cupboard in a different order, testing the algorithm's ability to handle grasp choices and infeasible placements under shelves.

Unpack Benchmark

Unpack problem: Objects initially placed on a tray-like surface must be unpacked and ordered into a cupboard. The challenge lies in selecting feasible grasps due to the proximity of objects and the restricted goal placements under a shelf, which disallow top grasps.

Swap Benchmark

Swap problem: A set of objects are initially placed either on the table or on higher support surfaces, and the goal is to swap their poses. The challenge arises from the fact that the goal poses of objects are already occupied, requiring the use of intermediary placements to achieve the desired configuration.

Access Benchmark

Access problem: The task involves moving a single object (red) to its goal placement while ensuring all other objects are returned to their initial positions. The challenge arises as each object blocks access to the next, requiring the robot to sequentially move blocking objects to temporary placements, position the red object, and then restore the moved objects in their original order.

Sort Benchmark

Sort problem: The task involves moving a set of objects from a shelf to one of two small tables based on their color. The challenge lies in placement sampling, as the planner must find feasible placements that allow all objects of the same color to fit on narrow surfaces. Additionally, the initial proximity of objects requires a specific order of movement. The problem is further complicated by the presence of blocking objects (orange) with unspecified goals on the tables.

Results

Detailed planning times of our algorithm on the Reorder, Unpack and Swap problems with and without feasibility prediction, compared to our previous work, averaged over 10 runs.

Result on the Reorder Problem

Reorder Problem

Result on the Unpack Problem

Unpack Problem

Result on the Swap Problem

Swap Problem

Detailed planning times of our algorithm on the Access and Sort problems with and without AFP-Net, averaged over 10 runs.

Result on the Access Problem

Access Problem

Result on the Sort Problem

Sort Problem

Visualizations

Plan execution visualizations for each problem.

Visualization of the solution to the Reorder Problem

Reorder Problem

Visualization of the solution to the Unpack Problem

Unpack Problem

Visualization of the solution to the Swap Problem

Swap Problem

Visualization of the solution to the Access Problem

Access Problem

Visualization of the solution to the Sort Problem

Sort Problem

BibTeX

          
            @inproceedings{ait2023simultaneous,
              title={Simultaneous action and grasp feasibility prediction for task and motion planning through multi-task learning},
              author={Ait Bouhsain, Smail and Alami, Rachid and Simeon, Thierry},
              booktitle={2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)},
              pages={2042--2048},
              year={2023},
              organization={IEEE}
            }