Towards Characterizing Feasibility of Edge Driven Split-Control in Bilateral Teleoperation of Robots

Human-centric bilateral teleoperation of robots over a network enables a human controller to perform assistive or corrective control to modify the state of a remotely placed robot. However, the timeliness of control depends on dynamics such as communication latency, bandwidth and quality/quantity of feedback information. In various specific use cases, inadequate compliance to communicate expeditious control could result in hazards. In this paper, we demonstrate the architecture of the split-control mechanism to enable edge-based control in bilateral teleoperation. Edge computing allows a robotic setup to benefit from the introduction of intelligent algorithms at the edge. We characterize the notion of feasibility of edge-based analytics by demonstrating two scenarios viz., a command controlled see-saw-like robot (with a low-complexity data/intelligence to capture the delay from cloud to edge); and a haptic glove controlled robotic hand (with an ML-based intelligence algorithm without incorporating the edge-to-cloud delay). We report a practical implementation of a 5G Mobile Edge Computing system that is leveraged to implement the specified split-control mechanism to detect and avoid certain unwanted events arising from the large latency between a robot and a remote human controller. Experimental results show that the proposed edge-based split-control system notably aids the human-centric control mechanism. We, then, introduce the notion of edge-fiability to capture the limits of edge-based prediction and control approach.