Course Websites

ECE 598 IBR - Info Bottlenecks in Robotics

Last offered Spring 2027

Official Description

Subject offerings of new and developing areas of knowledge in electrical and computer engineering intended to augment the existing curriculum. See Class Schedule or departmental course information for topics and prerequisites. Course Information: May be repeated in the same or separate terms if topics vary.

Section Description

How much must a robot remember, sense and communicate to complete a task? This course covers techniques for establishing such lower bounds and algorithms that construct policies, filters and encoders meeting or approaching them. Part I develops worst-case, set-valued techniques: quotients by indistinguishability (the Myhill-Nerode theorem, minimal sufficient information spaces), Hankel factorization, and covering and topological entropy arguments. These determine the smallest budget at which a task can be guaranteed. Part II develops average-case, measure-theoretic techniques: the minimal dimension of a Bayesian filter, one-shot rate-distortion theory, and directed information with sequential rate-distortion. These give trade-off curves between budget and performance. They come with algorithms for optimal coder-controllers. Examples come from micro- and nano-scale robots with severe budgets and from current problems in robotics: the camera resolution and frame rate a racing drone ne

Related Faculty

TitleSectionCRNTypeHoursTimesDaysLocationInstructor
Info Bottlenecks in RoboticsIBR69907LEC40930 - 1050 M W  4070 Electrical & Computer Eng Bldg Sayan Mitra