Our favourite on-screen robots move in a myriad of ways: from shape-shifting Transformers to the trundling R2-D2 and walking C-3PO of Star Wars fame. This diversity mirrors the real world, where robots are specially designed to manoeuvre often challenging environments.
Wheeled robots are typically selected for carting heavy loads across smooth terrain. However, wheels won’t get robots very far on rocky, uneven surfaces. Legged robots do much better here, mimicking human or animal movements to negotiate obstacles. And now, robotic experts are working towards a hybrid robot that combines the best of both worlds.
“To make a robot seamlessly traverse through terrain in the real world over even and uneven surfaces, we propose a new method for a hybrid wheeled-legged robot,” said Samuel Cheong, a Research Engineer at A*STAR’s Institute for Infocomm Research (I2R).
Hybrid robots have the potential to be used in complex physical environments like disaster zones, and perform high-risk operations on search and rescue missions, or safely inspect and dispose of bombs. According to Cheong, the main flaw in current hybrid robots is that they can’t seamlessly switch between rolling and walking modes.
“Such mode switching may cause the robot to stop for a while,” explained Cheong. “Such recalibration time could mean the difference between life and death in extreme situations.”
In their study, Cheong and colleagues propose a new sensor framework for next-generation wheeled-legged hybrid robots. The goal was to move away from classical leg odometry, where sensors estimate each leg’s position on the ground before taking the next step. The computing time for this step triggers the pause between rolling and walking modes.
“To bypass the need for switching between rolling and stepping, we used an Instantaneous Contact Point (ICP) as a replacement for the fixed contact point,” said Cheong, adding that the team used a combination of wheeled legs and cameras that map the three-dimensional topography of the ground in real-time to achieve smooth rolling-walking transitions. They also developed Hybrid Locomotion Odometry technology, which for the first time, enables robots to take a step forward while their other legs are rolling.
Their creation has been successfully integrated into a two-armed, four-legged robot called Pholus, enabling it to move nimbly and sense when it should side-step obstacles while travelling on its wheels. “This is like how humans walk along the road based on sensations from our legs and perception by looking ahead,” said Cheong, whose team is working towards smarter, autonomous and more versatile robots of the future.
The A*STAR-affiliated researchers contributing to this research are from the A*STAR Institute for Infocomm Research (I2R).