Fill ‘Er Up

Applications for autonomous robots are still limited, but with the current rate of advancement, we can expect to see more robots making deliveries, stocking shelves, and working in factories in the near future. One problem that has not yet been fully overcome, however, is determining how to power these devices. When the robots need to travel beyond the reach of a recharging station, they still need a way to keep their batteries topped off.

A team from the Skolkovo Institute of Science and Technology has recently put forth a solution to this problem that could keep autonomous robots running, reduce the time required to complete their tasks, and increase their range. Their idea was to create a robot, named MobileCharger, that can autonomously locate other robots that need a battery recharge, and power them back up while they are still on the go.

The key innovation in MobileCharger is the DeltaCharger system, which is able to position electrodes in three dimensions for making contact with the target robot’s recharging pads. Initially, an RGB depth camera locates the charging port of the target robot. The image data is passed to a convolutional neural network (CNN) that detects charging pads.

Computer vision alone has problems in detecting small misalignments when the contacts are in close proximity, so high-density pressure sensors are the primary means of ensuring a good contact. The tactile data received from these sensors is processed by a CNN to determine when proper contact has been achieved.

The DeltaCharger system was evaluated by the researchers, and it was found that it was able to measure the angle, and amount of vertical and horizontal electrode misalignment with 95%, 98%, and 86% accuracy, respectively. These results suggest that MobileCharger has the potential to recharge a wide variety of robots, with differing styles of charging ports.

Dzmitry Tsetseruko, one of the study authors, noted that future revisions of MobileCharger may be able to stay in the field indefinitely, by finding a sunny location to recharge with a solar panel when their own power supply starts to run low. The present device is still in the early stages of development, with work on collision avoidance, localization, and mapping algorithms on the development list before it is ready for use in the wild. If these enhancements make it into a future revision, MobileCharger could ultimately be a very useful tool that enables the furthering of autonomous robot applications.