State-of-the-art radiotherapy machines with integrated magnetic resonance (MR) imaging, known as MR-Linacs, provide the capability to track tumors in real time. This capability aids delivery of precise irradiation in the presence of patient motion, such as breathing, by adjusting the radiation beam. However, current solutions rely solely on geometric tracking without closing the loop by considering the actual endpoint-the delivered radiation (here called dose). Real-time dose-based adaptation within a single session remains highly challenging due to the immense dimensionality of the problem. To overcome this, we have developed a radiotherapy simulator and propose a novel reinforcement learning (RL)-based approach for real-time adaptive radiotherapy using 2D fluence, as a surrogate to 3D dose. To our knowledge, this is the first application of RL in real-time adaptive radiotherapy. Our in-silico experiments showed the feasibility of using RL to close the feedback loop, dynamically adapting to patient motion and minimizing discrepancies between delivered and intended dose in clinical cases. Our approach introduces a new treatment delivery paradigm, enabling delivery based on a reference fluence and motion without predefined machine settings.

We study the attitude estimation problem for an accelerated rigid body using gyros and accelerometers. The application in mind is that of a walking robot and particular attention is paid to the large and abrupt changes in accelerations that can be expected in such an environment. We propose a state estimation algorithm that fuses data from rate gyros and accelerometers to give long-term drift free attitude estimates. The algorithm does not use any local parameterization of the rigid body kinematics and can thus be used for a rigid body performing any kind of rotations. The algorithm is a combination of two non-standard, but in a sense linear, Kalman filters between which a trigger based switching takes place. The kinematics representation used makes it possible to construct a linear algorithm that can be shown to give convergent estimates for this nonlinear problem. The state estimator is evaluated in simulations demonstrating how the estimates are long term stable even in the presence of gyrodrift.

A novel method for offline scheduling is proposed where the periodic task model does not contain the usual task period parameter. This is possible because the scheduling is integrated with control theory. A slot length defines the granularity of schedule and the optimal control theory tells us what the best offline static cyclic schedule is with respect to the overall system performance. The optimal control theory used here is periodic linear quadratic control and should be comprehensible to the control engineer versed in state space theory. The scheduled resource is fairly divided among the controllers by paying respect to their needs when rejecting disturbances acting on the plant.

We study a nonlinear pitch and roll estimation problem for a 4-rigid body, aiming at a walking robot application. The approach taken is that sensor data from rate gyros and inclinometers are combined using a high-gain observer that can be proven to be exponentially convergent. The algorithm has successfully been evaluated experimentally during conditions resembling walking robot motion and has been compared with the more standard extended Kalman filter (EKF). It is shown that the much simpler high-gain observer performs slightly better than the EKF and that both algorithms provide small and bounded estimation errors.