Home research People General Info Seminars Resources Intranet
| Algorithms & Applcations Group | Home | Research | Publications | People | Resources | News
Motion Planning Benchmarks | Algorithms & Applications Group
Algorithms & Applications Group
Motion Planning Puzzles
(aka benchmarks)

The models contained on this page represent our attempt to try to start a collection of benchmark problems that can be used to compare various motion planning algorithms. As such, these models are available for public, non-commercial use provided that appropriate reference is made to the source/creator of the model. Also, if you have any models that you could contribute to this effort, please let us know and we will be happy to post them on this page.

Alpha Puzzle
Provided by Boris Yamrom, GE Corporate Research & Development Center

The alpha puzzle benchmark is a motion planning problem containing a narrow passage. The puzzle consists of two tubes, each twisted into an alpha shape; one tube is the obstacle and the other the moving object (robot). The objective is to separate the intertwined tubes.

In order for the problem to be solved a complex set of translation and orientation movements need to take place within the narrow passage. Computationally the narrow passage needs to be adequately mapped which is a difficult problem as this valid space is fairly difficult to generate samples in.

Animations:

Flange Problem
Provided by GE Corporate Research & Development Center

The flange benchmark is a motion planning problem containing a narrow passage. It is representative of a type of problem that might occur in a maintainabilty study of a mechicanical CAD design. The environment consists of a fixed rectangular part with a circular opening (the obstacle) and an elbow shaped curved pipe (the robot) that must be inserted into the opening in the obstacle.

This problem requires generating a sliding motion where the obstacle and robot are nearly touching with the robot twisted into or out of the obstacle. Finding valid motions in this constrained space is a difficult problem with many approaches developed to find these portions of the space.

Animations:

Box Folding Problem
Created by Guang Song, Parasol Lab, Texas A&M University

The box folding problem is a motion planning problem containing a narrow passage. The objective is to fold the articulated model into its final box shape. There are no external obstacles in the enviroment, but self-collision among the links (connected by revolute joints) must be avoided.

Animations:

Periscope Folding Problem
Created by Guang Song, Parasol Lab, Texas A&M University

The periscope folding problem is a motion planning problem containing a narrow passage. The objective is to fold the articulated model into its final periscope shape. There are no external obstacles in the enviroment, but self-collision among the links (connected by revolute joints) must be avoided.

Animations:

Pentomino Puzzle
Created by Ian Remmler, Parasol Lab, Texas A&M University

The Pentomino Puzzle environment was created to test disassembly planning methods. The puzzle is "solved" in its initial (assembled) configuration, and the goal is to disassemble the puzzle by moving each piece an arbitrary distance away from all other pieces.

Animations:

Bug-Trap
Created by Parasol Lab, Texas A&M University

The objective is to take the bug robot outside of the trap obstacle throught the narrow passage.

       

Hedgehog
Created by Vojta Vonasek at the Intelligent and Mobile Robotics Group, Department of Cybernetics, Czech Technical University

The Hedgehog environment contains a few very difficult narrow passages that the robot (spiked sphere) must pass through to escape the cage.

Serial Walls
Created by Parasol Lab, Texas A&M University

The Serial Walls environment consists of a set of walls (with openings) that the stick (the robot) has to traverse.

Animations:

S Tunnel
Created by Parasol Lab, Texas A&M University

The S tunnel environment consists of two elongated cubes. The obstacle is an elongated cube with a long narrow tunnel incisioned in it. The robot is another small cube.

Z Tunnel
Created by Parasol Lab, Texas A&M University

The Z tunnel environment contains an obstacle with a single narrow passage shaped like a Z. The robot is a small cube.

L Tunnel
Created by Parasol Lab, Texas A&M University

The L tunnel environment contains two L-shaped passages the robot (also L-shaped) must traverse.

Animations:

Bright Building
Created by Parasol Lab, Texas A&M University

The Bright Building environment contains the first 4 floors of HRBB. The building consists of many rooms, corridors, and stiarwells. Agents navigating in this environment must be able to extract paths quickly through the environment in order to perform certain behaviors.

Animations: