In iRaL, we care for a trustworthy collaborative autonomy in dynamic, interactive, and adversarial environments. We care for impact on safety- and time-critical applications that necessitate trustworthiness, such as:
- multi-target tracking, search and rescue, and persistent surveillance;
- safe navigation of self-driving vehicles in crowded cities, and unknown (extraterrestrial) environments;
- attack-susceptible satellite networks and flying multi-rotor vehicles.
We aim for a technological convergence between:
- “cyber” capabilities for a distributed intelligence, driven by online learning for perception, (inter)action prediction, planning, and control, and
- “physical” capabilities of novel self-reconfigurable aerospace systems.
We build on tools of learning, perception, control, as well as, of combinatorial, non-convex optimization, with the goal to:
- identify fundamental optimization limits;
- develop provably optimal algorithms;
- build novel robotic platforms;
- apply the research outcomes.
.
Vasileios Tzoumas 