Remote Control Cockroach

This insect-computer hybrid uses the natural flight and navigation systems of cockroaches to limit energy usage of search and rescue robots.

Nick Bild
3 years agoMachine Learning & AI
Insect-computer hybrid (📷: P. Tran-Ngoc et al.)

A swarm of tiny flying robots that can find humans in search and rescue missions has long been a dream of technology-minded emergency rescue personnel. Great strides have been made towards this goal in recent years, however, the energy required to support locomotion, and the computationally expensive autonomous control algorithms and associated sensors have kept this dream just out of reach.

A team led by researchers at Nanyang Technological University of Singapore has taken a new approach to the problem that eliminates the need to provide for locomotion, as well as many of the typical control systems. Their insect-computer hybrid system attaches a minimal set of sensors and processing units to the back of a Madagascar cockroach.

The backpack contains a Texas Instruments MSP432P4011 microcontroller as the main processing unit. An infrared camera, Bluetooth transceiver, CO2 sensor, temperature/humidity sensor, and inertial measurement unit are also onboard. Additionally, a 2 megabyte flash memory is included to record activities during operation. The entire backpack weighs 5.5 grams, which is well within the insects’ maximum payload of approximately 15 grams. The system can run for an hour or more before needing a recharge.

The insect provides flight capabilities for the hybrid system. The innate obstacle avoidance and navigation skills of the cockroaches are also leveraged to reduce the complexities of the tasks required of the backpack. The infrared camera, coupled with a machine learning algorithm is capable of locating humans. To direct the insects’ flightpath, the team attached electrodes from the backpack to their cerci to provide electrical stimulation. When the left cercus is stimulated, the insect will fly to the right, and conversely, when the right cercus is stimulated the insect will fly to the left.

Overall, the human detection algorithm achieved an 87% average accuracy; when the target is within 1.5 meters, the accuracy was found to average 90%.

While this insect-computer hybrid system is still in the proof of concept stage, it represents an important step forward for low-power, mini-scale, autonomous search and rescue robotics. The tiny robots have shown that they are capable of operating autonomously in real-world environments and avoiding obstacles in search of humans in distress. The team is working to add an accurate on-board localization system for tracking the position of the insect in real time to speed response times after the system locates a victim.

Nick Bild
R&D, creativity, and building the next big thing you never knew you wanted are my specialties.
Latest articles
Sponsored articles
Related articles
Latest articles
Read more
Related articles