Active Bionic Antennas for Object Detection and 3D Localization with Ego-Motion Cancellation
Team
- E/19/295 Janitha Dilshan — e19295@eng.pdn.ac.lk
- E/19/300 Asela Hemantha — e19300@eng.pdn.ac.lk
- E/19/495 Dulshan Chamuditha — e19495@eng.pdn.ac.lk
Supervisors
- Dr. Nalin Harischandra — nalin@eng.pdn.ac.lk
- Dr. Isuru Nawinne — isurunawinne@eng.pdn.ac.lk
Table of Contents
- Abstract
- Related Works
- Methodology
- System Architecture and Implementation
- Results and Analysis
- Conclusion
- Publications
- Links
Abstract
Inspired by the tactile sensing mechanisms of insects and mammals, this project presents a biologically-inspired bionic antenna system capable of 3D object detection and localization. The system cancels ego-motion artifacts using Echo State Networks (ESNs), enabling robust performance in vision-impaired environments. Key applications include autonomous robotics, search and rescue, and industrial inspection.
Related Works
- Previous tactile sensors like robotic whiskers and flexible antennae showed promise but struggled with motion-induced noise.
- Modern research applies reservoir computing and deep learning for proprioceptive signal processing and localization.
- Our system extends this by incorporating real-time motion cancellation using ESNs.
Methodology
- Design of flexible, servo-actuated antennae mimicking insect behavior.
- Integration of IMU sensors (MPU6050) and vibration data acquisition at ~130Hz.
- Application of Butterworth and band-pass filters to preprocess raw vibration signals.
- Offline training of Echo State Networks (ESNs) to model and cancel self-induced motion noise.
System Architecture and Implementation
- Mobile Platform: 4WD robot car.
- Processing Unit: NVIDIA Jetson Nano for ESN inference.
- Sensor Module:
- flexible bionic antennae with servo motion.
- MPU6050 IMUs to capture vibration and acceleration.
- Software Stack:
- Real-time ESN-based ego-motion cancellation.
Results and Analysis
- Reduced False Positives: Ego-motion noise cancellation improved detection accuracy.
- Improved Localization: Antenna-based mapping functioned well in degraded environments (e.g., low light, dust).
- High Tactile Sensitivity: System responded reliably to light object contact, suitable for fragile environment exploration.
- Limitations:
- Antenna accuracy depends on servo resolution and noise.
- Slight inconsistency due to mechanical vibration in low-quality servo motors.
Conclusion
This project demonstrates the first integration of bionic antennae with ESN-based ego-motion cancellation into a mobile robotic system . The platform can effectively navigate environments unsuitable for visual or LiDAR sensing, opening up new domains for tactile-based autonomy in robotics.
Publications
Links
- 🔬 Project Repository
- 🌐 Project Web Page
- 🏫 Department of Computer Engineering
- 🎓 University of Peradeniya