Bio-inspired vision based robot control using featureless estimations of time-to-contact

Area of Science:

• Robotics
• Computer Vision
• Control Systems

Background:

• Vision-based dynamic behaviors in nature inspire robotic applications like landing and obstacle avoidance.
• Time-to-contact (TTC) estimation is crucial for robotic navigation but often computationally intensive.
• Existing TTC methods require feature extraction and tracking, limiting their use in resource-constrained robots.

Purpose of the Study:

• To develop a computationally efficient, featureless method for TTC estimation.
• To extend TTC estimation to include angular velocities and enhance accuracy using Kalman filtering.
• To design an error-based controller with gain scheduling for mobile robot motion control.

Main Methods:

• Implemented a featureless TTC estimation technique.
• Extended the method to incorporate angular velocities.
• Applied Kalman filtering to improve TTC estimation accuracy.
• Developed a gain-scheduled, error-based controller for mobile robots.

Main Results:

• The featureless TTC estimation method was successfully extended and improved.
• Kalman filtering enhanced the precision of TTC estimation.
• The developed controller effectively managed mobile robot motion.
• Experimental validation on a low-cost embedded system demonstrated successful robot docking.

Conclusions:

• The proposed featureless TTC estimation and control methods are effective for computation-constrained robots.
• This approach enables agile locomotion tasks like landing, docking, and navigation for miniature robots.
• The methods are suitable for real-time applications on embedded systems with limited computational power.