Simultaneous Localization of Biobotic Insects using Inertial Data and Encounter Information

Area of Science:

• Robotics
• Bio-inspired engineering
• Insect locomotion

Background:

• Bioelectrical stimulation enables control over Madagascar hissing cockroach (Gromphadorhina portentosa) locomotion.
• This control presents opportunities for using insects in centimeter-scale environmental exploration, such as urban disaster rubble.

Purpose of the Study:

• To develop and evaluate a machine learning-based inertial navigation system for localizing groups of G. portentosa.
• To enhance trajectory reconstruction accuracy by incorporating inter-agent encounter data.

Main Methods:

• Implemented an inertial navigation system utilizing machine learning modules.
• Equipped G. portentosa with thorax-mounted inertial measurement units.
• Integrated agent encounter data as signals of opportunity to improve localization.

Main Results:

• Achieved localization of multiple G. portentosa agents on a planar surface under laboratory conditions.
• Trajectory reconstruction accuracy improved by 16% using encounter information.
• A further 27% improvement in accuracy was observed with a heuristic for optimal speed-scaling.

Conclusions:

• The developed navigation system effectively localizes G. portentosa.
• Utilizing agent encounters significantly enhances navigation accuracy for insect-based exploration.
• This technology holds promise for remote sensing in challenging, confined environments.