Separating Intrinsic and Extrinsic Responses of Whisker Sensors Using Accelerometer

Rodents and Felidae whiskers are highly sensitive, detecting extrinsic inputs such as airflow or contact and intrinsic inputs such as base vibrations or self-induced motion. Building effective artificial whisker sensors faces a challenge due to the intricate coupling of responses at the whisker base. There is a research gap in understanding whisker sensors’ responses to intrinsic and extrinsic inputs. To address this, we propose two methods, using base acceleration as a reference input: 1) employing frequency-domain adaptive filtering (FDAF) and 2) introducing the base vibration response model (BVRM) that mathematically represents the whisker sensor’s behavior to base vibrations or self-induced motion. Validation of FDAF and BVRM is conducted through simulation and experimentation. The BVRM excels in both simulation and experiment, demonstrating a signal-to-noise ratio (SNR) of 35.20, slightly outperforming the laboriously tuned partitioned constrained FDAF with an SNR of 34.96, despite FDAFs slower convergence and poorer performance in experiments. In addition, BVRM can be useful in filtering sensor responses for independent use cases, such as terrain identification, flow sensing, and surface profile identification. By separating responses to extrinsic and intrinsic inputs without discarding either, whisker sensors become more versatile and multipurpose.