Brain-mimicking ‘neuromorphic’ computer chips for robot insects

Cornell engineers are developing new programming that will make RoboBees more autonomous and adaptable to complex environments (Credit: Harvard Microrobotics Lab) 

Instead of binary code, neuromorphic chips process spikes and combinations of electrical current, in a similar way to how neurons fire in the brain. ‘Event-based’ sensing and control algorithms created by the researchers would allow robot insects to react to the unpredictable scenarios that play out all the time in the real world. And as they require significantly less power than traditional processors, more computation can be packed into smaller payloads.

The Cornell team is working with the Harvard Microrobotics Laboratory, creators of the RoboBee microrobot. Currently, RoboBee remains tethered to a power source while operating, but the team behind it is looking to remove that constraint. According to the Cornell engineers, its neuromorphic algorithms could help make RoboBee more autonomous and adaptable to complex environments, with little additional weight.