Scientists from Singapore have developed ‘mini-brains’ that can allow robots to feel the equivalent of pain and determine when they should start to repair themselves.
These processing units are embedded into the surface, or the robotic equivalent of skin, allowing the damage to be detected and analysed where it occurs.
This is a different approach from current systems, which usually involve a robot fitted out with various sensors, such as cameras, microphones and touch sensors.
The data picked up by these sensors is not analysed at source, but sent to a centralised processing unit.
This usually requires extensive wiring for the information to travel along, and can involve a significant delay between the sensor receiving the data and the robot acting upon it.
The new system, developed by scientists at the Nanyang Technological University (NTU Singapore), embeds AI into a network of sensor nodes.
This is connected to the series of ‘mini-brain’ processing units known as memtransistors, which are less powerful than a single central processor, but allow for localised learning.
These memtransistors are able to memorise and process information, acting like synapses and pain receptors.
System can reduce costs and improve response times
The system reduces the need for wiring – which can be vulnerable to damage – and also reduces response time by between five and 10 times compared to conventional robots.
It was able to respond to ‘pain’, registered through sensors in the form of external pressure that crushes or cuts, in real time.
When its material was cut or crushed, the robot would lose a measure of functionality.
The team combined the sensory system with an ion-gel material that was able to self-heal if this happened.
Essentially, the molecules of the gel knit together, allowing the robot to initiate minor repairs without human help.
Rohit Abraham John, lead author of the research, said that their robot was able to repeatedly repair itself when ‘injured’, in a similar way to biological skin healing itself from a minor injury.
The fact that it could continue to operate effectively after sustaining damage could therefore lead to maintenance savings in a real-world setting.
Co-lead author Nripan Mathews said that their work was partly inspired by human neuro-biological functions.
“While still at a prototype stage, our findings have laid down important frameworks for the field, pointing the way forward for researchers to tackle these challenges,” he added.
Today’s news was brought to you by TD SYNNEX – the UK’s number one solutions distributor.
