ARTOR (Autonomous Rough Terrain Outdoor Robot), wheeled ground robot

[Airicist]

Website - artor.ethz.ch

This project aims at the development of techniques for reliable autonomous navigation of a wheeled robot in rough, outdoor terrain. The robot must be able to navigate and localize itself in unknown, challenging environments without using global position sensors (such as GPS). Leaving flat and well-structured environments, such as streets or office rooms, comes along with a series of challenges for navigation. The terrain not only becomes three-dimensional, but also exhibits various different surfaces (vegetation, gravel, sand, rocks, etc.).

For reliable autonomous navigation in such terrain, the robot must first localize itself in six dimensions and build a three-dimensional map of the environment, based on sensor data. Furthermore, the traversability of individual parts of the surroundings has to be determined, in order to be able to plan a safe path leading the robot towards the desired location. Computation of this path is not only based on the need to avoid obstacles, but also on the shape (steepness) and traversability of different parts of the terrain.

[Airicist]

Teach And Repeat in Dynamic Environments

Published on Apr 8, 2013

This video presents preliminary results for Teach & Repeat task where the robot is manually driven around the environment and is asked to repeat the path.

The robot used in this case was ARTOR developed at the Autonomous Systems Lab in Zurich. A teach 1.3 km long path was recorded during the night and then repeated autonomously four times in different lighting and weather conditions. During the repeat phase, the algorithm shows robustness against multiple dynamic elements (i.e., pedestrians, bikes, cars, trucks, etc.).