The Mode Sensing Hypothesis

These studies have revealed a fundamental design principle in biological sensorimotor architectures: sensing is most effective when it is tuned to the dynamics and actuation of the body it serves. Using a physics-based model of blowfly flight together with electrophysiological measurements of visual neurons, we show that the fly’s motion vision system is aligned with the dynamically important directions of its flight dynamics, maximizing the transfer of signal energy from disturbances and control inputs to sensory outputs (Humbert et al.). Building on this discovery, this principle is transitioned to a synthesis framework for designing distributed sensing systems that match sensing characteristics to system dynamics and actuation (Turin et al.). By optimizing the jointly controllable and observable directions encoded by sensors using control-theoretic metrics such as Hankel singular values, the approach minimizes the control energy required for stabilization and improves robustness to uncertainty. Together, the results establish a new paradigm for bio-inspired robotics and autonomous vehicles in which sensing, actuation, and dynamics are co-designed to enable agile, efficient control with reduced computational overhead.