Robot Navigation Strategy Based on Honeybee Learning Flights Proposed in Nature Paper

A research paper titled 'Efficient robot navigation inspired by honeybee learning flights' was published in Nature. The paper proposes a navigation strategy named Bee-Nav inspired by the visual learning flights of honeybees. Bee-Nav uses a tiny neural network trained during robotic learning flights to map omnidirectional images to a home vector based on path integration.

Honeybees perform one to several short learning flights before departing on longer exploration or foraging trips. The learned homing area, or LHA, is the area circumscribing the learning flight trajectories within which the neural network can estimate the home location. 00% of the total flight area.

24%. 6%.

4%. The learning flight trajectory fits in a 10-m-radius circle around the home location. 5 m of home for 100% of 30–110-m flights.

5 m of home for 70% of 200–600-m flights in windy conditions using a 42-kB neural network. 3-kB attention neural network.

Visual homing success rate of Bee-Nav was 100% within the LHA radius in ten different simulated forests. 5 times the LHA radius depending on the environment. Within the LHA the angular errors of the network were smaller than about 40°.

Visual homing is generally successful if angular errors stay below 90°. Bee-Nav visual homing performed substantially better than a snapshot-based approach and a perfect memory approach that stored all learning flight images.

Path integration for the robot is based on velocity from optical flow and laser-based height measurements and heading based on gyro integration. The view memory consists of a small feedforward convolutional neural network that maps an unwrapped omnidirectional image to a home vector. A recent robotics study used a 9-kB neural network for visual homing up to 18 m outdoors.

The state of the art prior to this paper was a tiny flying robot using 500 kB of memory for navigating in a 4 × 5-m area. Honeybees can navigate up to several kilometres from their hive.