Monocular distance estimation with optical flow maneuvers and efference copies: a stability-based strategy

Area of Science:

• Robotics
• Computer Vision
• Insect Navigation

Background:

• Optical flow is crucial for flying insect and robot navigation.
• Accurate distance estimation is a challenge for optical flow control systems.
• Existing methods struggle to derive distance from the velocity-to-distance ratio provided by optical flow.

Purpose of the Study:

• To propose a novel, stability-based strategy for monocular distance estimation.
• To enable robots to estimate distance using optical flow maneuvers and efference copies.
• To overcome limitations in current optical flow-based navigation systems.

Main Methods:

• Analytical derivation of control loop stability based on distance.
• Utilizing self-induced oscillations within a control loop to infer distance.
• Implementing and testing the strategy on simulations and a Parrot AR drone 2.0.

Main Results:

• Control loop stability is shown to depend on the distance to the approached surface.
• Self-induced oscillations at close distances reliably indicate proximity.
• The strategy robustly estimates distance during hover and landing maneuvers, unaffected by wind.

Conclusions:

• The proposed stability-based strategy offers a robust method for monocular distance estimation.
• Robots can detect oscillations to gauge distance, enhancing navigation capabilities.
• The method supports triggered landings, hover-based distance estimation, and adaptive gain control for continuous landing.