I attended the 2025 American Control Conference in Denver. Two of my students attended and presented papers (Milad, left, and Mohammad, right). This was their first time attending an academic conference!

The paper with Milad and myself was titled “Spacecraft attitude control under reaction wheel constraints using control Lyapunov and control barrier functions”. In short, Milad proposed an approach for rest-to-rest maneuvers of satellites using reaction wheels for attitude control. Instead of traditional optimal control approaches that are computationally intensive and must be carried out offline, Milad’s approach uses control Lyapunov functions (CLF) and control barrier functions (CBF) to produce a closed-loop control strategy that can adapt to sudden changes in constraints. Another contribution of the work is to resolve an issue of chattering that is common with CLF-CBF approaches by introducing a novel regularization term in the associated quadratic program.

The paper with Mohammad and myself was joint work with Rifat Sipahi, a colleague at Northeastern. The paper was titled “Stealthy optimal range-sensor placement for target localization”. When using range-only sensors, the error in localizing a target is a function of the number of sensors and their configuration in space. In general, when the sensors are aligned, localization is poor. The quality of sensing can be quantified using the notion Fisher information. Given a target, we can ask: how should we arrange a set of sensors to best localize it (the Fisher information we have about the target is large)? Similarly, given a set of targets, we can ask: where should we place ourselves so that we are as stealthy as possible (the Fisher information the targets have about us is small). In this paper, we look at the combination of both problems: given a group of targets, how should a group of agents arrange themselves so that they are simultaneously able to localize all the targets while remaining stealthy. The problem is difficult in general, but we solve the special case of two agents and two targets, and present scalable performance bounds for the case of arbitrarily many agents and two targets.

In other news, I was really impressed by the recent progress in robotics, particularly walking/humanoid robots. Below is a short video of a demo by Unitree Robotics of a remote-controlled humanoid robot. Robust, natural-looking walking and running. Very impressive!

Thanks and congrats to the organizing committee and everybody else that came together to make ACC a great success once again!