Robot could guide humans through areas of low visibility
Sheffield Hallam University is taking part in a multi-disciplinary project that aims to develop a robot capable of guiding humans in areas of low visibility.
It is hoped that the EPSRC-funded Reins project, which involves King’s College London and Thales, will lead to a semi-autonomous mobile robot with sensory capabilities that can be shared with humans.
The new robot could be used as a tool by fire fighters and, according to Sheffield Hallam, be adapted further for people who are blind or visually impaired.
Key to the robot-human interaction are reins or lead-like tethers that provide communication between the robot and human, and visa versa.
The robot would provide haptic feedback to the human handler in low-visibility conditions, while a human operator could provide instructions to the robot via the reins — a concept based on the relationship between a rider and horse.
Three types of reins have been proposed: flexible, stiff (inspired by the lead for guide dogs) and wireless (based on the Wii handset). Common to all designs will be vibrations or waves that propagate through the reins.
According to Dr Jacques Penders, head of the Centre for Automation and Robotics Research (CARR) at Sheffield Hallam, the robot itself is likely to be slightly larger than a laptop and will incorporate a flexible bumper as its main sensor.
For fire fighters, swift movement can be critical in buildings filled with smoke. The Reins project seeks to improve robotic perception by creating a mechanism for impedance filtering so it can make a judgement on the delays that could be caused by unseen objects. Algorithms will further expand the knowledge available in tactile perception for mobile robots.
‘If you push the robot toward an empty box the robot will push it away, but if it meets a heavy obstacle the robot will generate some pressure to that obstacle,’ said Penders. ‘Whether it can move the obstacle or not is an indication for the user of what he or she can expect.’
By taking this approach, the robot will overcome the shortcomings of robots that rely on range finding and proximity sensing to navigate a cluttered environment.
Another key challenge will be to decide whether information should be encoded explicitly or remain implicit.
Penders said that, with the flexible rein, information would have to be explicit as the pull of the rein would classify the order to be executed. ‘You really have to define what a certain signal will mean,’ he said.
He added that the stiff-reined robot, which is being worked on at Sheffield Hallam, will likely provide implicit feedback.
Ultimately, the project hopes to create a machine that a human can trust.
‘Regardless of flexible or stiff reins, can it [the robot] create some confidence for the human using it?’ said Penders. ‘If you have no visual feedback you are losing a lot of information. Everyone within the consortium is aware of the risks you take there.’
The £430,000 project, which incorporates expertise from the fields of design, engineering, robotics and communication, is receiving input from South Yorkshire Fire and Rescue and the Guide Dogs for the Blind Association.
It builds on two robotic projects carried out by CARR, namely Guardians and Viewfinders.
Guardians are a ’swarm’ of autonomous robots that can navigate and search urban areas such as warehouses and factories. Viewfinders autonomously navigate through and inspect an area but human operators can monitor their operations, as well as control their movements if needed.