Antenna-Based Tactile Sensing for High-Speed Wall Following

Myriad creatures rely on compliant tactile arrays for locomotion control, mapping, obstacle avoidance and object recognition. Cockroach antennae are complex sensory structures whose mechanical properties are well tuned for certain behavioral tasks. Our laboratory is “reverse engineering” the neural controller for cockroach wall following to better understand sensorimotor integration in nature. In addition, we are building tactile sensors, inspired by their biological analogs.

Biologically Inspired Tactile Sensing

Like their biological analogs, robotic tactile sensors should enable their host to negotiate cluttered environments in low light. These multifunctional, light weight, low power, “quiet” sensors complement existing proximity sensors, particularly in low-light, tight spaces with highly polished surfaces and high air- or water-particle content, where modalities such as infrared, sonar, vision and lasers fail. Inspired by the cockroach antennae we are developing the next generation tactile sensors.

Antenna-based Control of the Lateral Leg Spring (LLS) Model of Cockroach Locomotion

The Lateral Leg Spring (LLS) model was developed by Schmitt and Holmes to model the horizontal-plane dynamics of a running cockroach. The model captures several salient features of real insect locomotion, and demonstrates that horizontal plane locomotion can be passively stabilized by a well-tuned mechanical system, thus requiring minimal neural reflexes. We treat the LLS as a “plant model” and biologically inspired control law that enables the model to follow along a virtual wall, much like antenna-based wall following in cockroaches.

Task-Level Control of Wall Following in Cockroaches

The American cockroach, Periplaneta americana, is reported to follow walls at up to 25 turns/s. During high-speed wall following, a cockroach holds its antenna relatively still at the base while the flagellum bends in response to upcoming protrusions. We developed a simple mechanosensory model for the task-level dynamics of wall following. In the model a torsional, mass-damper system describes the cockroach’s turning dynamics, and a simplified antenna measures distance from the cockroach’s centerline to a wall. Nyquist and root-locus analyses predict that stabilizing neural feedback requires both proportional feedback (“P”, difference between the actual and desired distance to wall) and derivative feedback (“D”, velocity of wall convergence) information from the antenna. These predictions led to electrophysiology in the antennal nerve in which we discovered “P” and “D” correlates.


Effects of Passive Antennal Reconfiguration During Wall Following in Cockroaches

We discovered that the passive antennal flagellum of cockroach Periplaneta americana can assume two principal mechanical states, such that the tip is either projecting backward or forward. Using a combination of behavioral and robotic experiments, we demonstrate that a switch in the antenna’s state is mediated via the passive interactions between the sensor and its environment, and this switch strongly influences wall-tracking control. When the tip of the antenna is projected backward, the animals maintain greater body-to-wall distance with fewer body collisions and less leg–wall contact than when the tip is projecting forward. We performed laser ablation of chemo-mechanosensory hairs and added artificial hairs to a robotic antenna through and showed that distally pointing mechanosensory hairs at the tip of the antenna mediate the switch in state by interlocking with asperities in the wall surface. Antennal hairs, once thought to only play a role in sensing, are sufficient for mechanically reconfiguring the state of the entire antenna when coupled with forward motion.