Robots that mimic the collective migration of biological cells are described in a paper published in Nature. The study describes an alternative approach to developing large-scale robotic systems with pre-determined behaviour that may be more scalable and robust than some current bio-inspired systems.
Modular or swarming robotic systems have been designed to mimic biological behaviours, such as self-assembly, healing and transport. Most systems require centralized control or have complex designs that limit the capabilities and scalability of the system. Hod Lipson and colleagues design a minimalistic system to reduce these constraints. Their robots are made up of simple disc-shaped ‘particles’, which they suggest are easier to manufacture and maintain in numbers than more complicated robots. Single robotic particles are not capable of movement; they just expand and contract like a camera aperture. However, they can loosely aggregate and be programmed to oscillate in an offset pattern in response to a signal gradient (such as light), which enables them to move collectively towards the stimulus.
Locomotion, object transport and movement towards a light stimulus are demonstrated in up to 25 physical robotic particles, and the authors show the system’s scalability in simulations with up to 100,000 particles. The system is also capable of carrying ‘dead weight’, with simulations predicting that it could maintain locomotion with 20% of the particles malfunctioning. The authors conclude that their robots display scalable control and robustness not seen in conventional robotic systems, in which sometimes the loss of a single component can lead to system failure.
Machine learning: Model identifies three biomarkers associated with COVID-19 mortalityNature Machine Intelligence
Planetary science: Life under a hydrogen atmosphereNature Astronomy