EPSRC Grant on Novel Visuo-Control Architectures for Agile Robot Exploration

News Image

03 November 2015

More News...
A new four year £1.6M project involving Bristol and Manchester Universities aims to explore vision and control algorithms for novel vision sensors and how these can be used in agile robot exploration.

Current robots that explore the environment are often slow and burdened with complex planning and control tasks. Vision is one of the primary senses that can help enable better autonomous agile robots, however, visual information processing is notoriously difficult especially at speeds required for fast moving robots. In particular, in situations such as micro air vehicles (MAVs) that require low weight, power dissipation and cost. Conventional hardware and algorithms are not up to the task. For example many calculations done on a Graphics Processing Unit (GPU) are expensive in terms of energy expenditure and could be wasteful as the entire image, which may contain many non relevant things, needs processing.

The new project funded by the Engineering and Physical Sciences Research Council (EPSRC) aims to tightly integrate novel sensing and processing hardware, together with computer vision, navigation and control algorithms, to enable the next generation of autonomous robots.

At Bristol, we will investigate how Computer Vision algorithms change when every pixel has its own processor; how processing at source helps to reduce sending irrelevant information through the visual pipeline; and how these features benefit a tighter integration with the control architecture. Aiming to demonstrate better and more agile operation of autonomous flying robots, the project involves various aerospace companies from the UK as well as academic collaborators overseas.

The novel vision sensors are a result of a long term research initiative at the University of Manchester while the Bristol's team contribution is based on our extensive work on real-time visual mapping, tracking and flight control and systems integration. One unique feature of this project is that complete sensor chipsets will be offered to promote development by external interested parties.

The work is lead at Bristol by Drs Walterio Mayol-Cuevas and Tom Richardson from the Departments of Computer Science and Aerospace Engineering, respectively.