Surrey Rover Autonomy Software & Hardware Testbed (SMART) for ProVisG

The AI and Autonomy research group at Surrey Space Centre is a partner of the ProVisG consortium who specializes in developing autonomous software for planetary rovers. The autonomy software consists of advanced functions achieved by implementing AI algorithms that enable rovers to operate in unknown planetary environment and conduct tasks autonomously including mission planning, localisation and mapping, object detection, path planning, and drilling, etc.

This group has developed the SMART system which supports testing and validation of new algorithms and autonomy functions for planetary rover research such as the work done within the ProVisG. The SMART rovers include a modified Surface Mobility Platform (see its picture on the left) and Pioneer 3AT. Their software system use the middleware based on Robot Operating System (ROS) framework that integrates different autonomy functions together with basic actuation and sensory functions on-board its rover. This is achieved by breaking the processing of raw sensor data into various stages that are pipelined to the integrated autonomy functions. The outcome of each processing stage is data products that are conformed from raw sensor data. A prominent feature of the middleware is that it enables each of the autonomy functions to be integrated and executed independently from other functions thus providing a modularised solution. Different tasks can also be launched concurrently and executed asynchronously while all pre-processed and processed data are shared between running tasks.

Up to the present day, the SMART system offers unique functions including the QujoteExpress mission planner, planetary monocular SLAM (PM-SLAM), monocular rock detection (RockMono), Lite-D* path planner (see its video demo at the bottom) as well as integration with a novel planetary drill payload. Among them, the PM-SLAM and RockMono will be particularly tested in the upcoming ProVisG field trial thereby generating advanced vision processing products.

Experimental tests using the SMART rover: