Chameleon 2018-06-20T14:50:38+00:00

Chameleon was a project conducted in partnership with SCISYS UK to investigate terrain-adaptive robotic navigation algorithms that could be used to enable long-range planetary exploration. This included the development of both established and novel algorithms, as well as validation on representative field data. This project was funded under the UK Space Agency’s CREST-2 Initiative.

Field Trials

The main highlight of our data collection efforts was a deployment with SCISYS to the Atacama Desert in September 2014. The vast scale of the test site was breathtaking. During the deployment, we conducted integrated system tests and gathered a large amount of data suitable for algorithm development.

From left to right: Dr. Mark Woods (SCISYS), Dr. Andy Shaw, SOLO (Rover, SCISYS), Dr. Iain Wallace (SCISYS), Dr. Chi Tong (Oxford), Mateusz Malinowski (SCISYS). Photo courtesy of Mark Woods.

Examples of terrain types of interest (from left to right): fine, lake, mixed, shale, and boulder field.

In addition to the adventure, the deployment provided valuable hands-on experience in operational procedures suitable for the field, and numerous first-hand encounters with the difficult of assessing terrain characteristics from appearance alone. Further details about the field deployment can be found in the field trial blog.

Scientific Contributions

Preliminary results using the data collected in the Atacama shows that the Experience-Based Navigation paradigm remains reliable for unstructured, planetary terrain. In addition, we used a copy of the space rover to develop the novel concept of perception schedules for energy-efficient path following. This study explored the idea of reducing energy consumption while following a path by turning off the main localisation subsystem and switching to a lower-powered, less-accurate odometry source at appropriate times. In effect, our goal was to activate the perception system only as required to maintain the rover within a given margin around the predetermined path. We were able to demonstrate a 12% reduction in total energy consumption through this perception scheduling method. This is of great interest for space exploration, as energy consumption is a significant restriction for long-range navigation. We continue to process the data and use it for further development.

Latest News

September 2016