To make the cooperation within a physical human-robot team as efficient as possible, the team members must be able to co-adapt.
We developed and evaluated a robot that adapts to a human, using an adaptation strategy, in such a way as to guide the co-adaptation to have a positive effect on human task contribution and team performance.
A novel adaptive control algorithm for the robot was designed, estimating and adapting to the human control, using a Nash equilibrium to compute the robot's control inputs. Stability of the controller was theoretically proven, and validated using physical tests.
Two robot adaptation strategies, positive and negative reinforcement, were compared in an experiment in which 18 participants participated. The negative reinforcement adaptation strategy provides assistance on an assist-as-needed basis, whereas the positive reinforcement strategy is designed to intrinsically motivate humans to contribute to the control task.
Results show a significant increase in performance in the negative reinforcement adaptation strategy compared to the positive reinforcement adaptation strategy, whereas both conditions show a significant increase in performance compared to manual control. Results additionally show a significant decrease in both estimated (by the robot) and perceived (by the human) control share in the negative reinforcement adaptation strategy compared to the positive reinforcement adaptation strategy.
In conclusion, to guide the co-adaptation to both increase performance and engage humans to actively contribute to a control task, a robot should be designed to adapt using a positive adaptation strategy.

In this thesis, we consider the problem of controlling multiple autonomous vehicles in a highway scenario, via MPC. By iteratively solving a motion planning OCP, MPC is perfectly suited for unknown dynamic environments, while optimally computing path and vehicle inputs. Moreover, MPC can ensure the satisfaction of collision avoidance constraints, a prerequisite for safe automated driving.

The collision avoidance constraints render the OCP non-convex. This thesis tackles this non-convexity by either designing nonlinear MPC controllers, or by convexifying these non-convex constraints.

Moreover, control of a large, networked system of automated vehicles is achieved by designing local, subsystem-based controllers. We analyse three different algorithms to distribute the plantwide OCP. All controllers are subjected to an objective analysis and compared to see which is the most efficient and most practical to implement. Centralized MPC is used as benchmark, since this gives the plantwide optimal solution. The first decomposed algorithm is decentralized MPC, where subsystems communicate a single time every MPC iteration and compute their new trajectory based on the previously communicated trajectory of neighboring subsystems. The second method is based on sub-optimal cooperative distributed MPC. Here, vehicles perform multiple sub-optimal iterations of a Gauss-Jacobi type distributed optimization. For the last method, based on a Generalized Potential Game, the vehicles sequentially solve and communicate the solution of their local OCP in order to find an $\epsilon$-Nash Equilibrium. By relying on additional constraints or fixed ordering among vehicles, all three controllers are able to recursively feasible compute their own trajectory while avoiding other vehicles.

The distributed controllers are assessed in two different scenarios, using three different criteria, i.e., the overall effectiveness of the controller, the local effectiveness of the controller and the progress made, by each vehicle in the simulation. The first criteria gives an indication of the level of cooperation among vehicles, the second shows the individual satisfaction of each vehicle with respect to its reference, and the last represents the overall progress each vehicle has made in the highway simulation.