High-performing control of complex dynamics, such as robots, requries to execute compute-intensive optimization at a remotely connected workstation that communicates with the robots in real time, receiving sensory inputs and sending control and navigation commands. While several control frameworks have been studied to satisfy safety requirements while maximizing performance, such as Model Predictive Control (MPC) and Control Barrier Function (CBF), standard schemes neglect the impact of nonideal wireless network on concurrent performance and safety requirements. This may lead to irreversible consequences, such as the robot crashing into an obstacle due to a blackout or network congestion that breaks sync between controller and robot.

In this thesis, you will develop a robot control scheme over WiFi that combines remote MPC and CBF with network modeling that explicitly considers wireless constraints such as latency, limited channel capacity, bandwidth, and noise. Development of algorithms will be complemented by simulations in MATLAB/Simulink and, possibly, experiments on a physical Segway-like balancing robot to validate the approach.