BioSoftRob - Design and development of bio-inspired representations and learning algorithms for modular simulated soft robots

BioSoftRob is a research project co-funded by the Oslo Metropolitan University (Norway), the University of Trieste (Italy), and Østfold University College (Norway).

Future robots will be more autonomous and will form robotic ecosystems that can react to changes in the environment they operate in. A key enabling factor for this scenario is modularity, that should regard both the body and the brain of the robot. Current state-of-the-art controllers (i.e., brains) for robots that are physically modular are not suitable for exploiting modularity.

The goal of the project is to explore the use of different paradigms for robot brains based on bioinspired Artificial Neural Networks (ANNs), e.g., Spiking Neural Networks (SNNs), represented in a way that is both amenable to optimization and modular by-design. SNNs are promising because they may give richer dynamics, due to the modelling of membrane potential in neurons, and because they can incorporate auto-adaptation and learning mechanisms. The work will consider the concrete case of Voxel-based Soft Robots (VSRs), a kind of robots composed of many simple soft blocks that are intrinsically modular in the body and allow for great freedom in the design of all of their subsystems (body, brain, and sensory apparatus).

The main outcome of the project will consist in advancing the state-of-the-art on representations and optimization algorithms that allow to embody ANNs and SNNs inside VSR blocks, such that disassembly and reassembly of robots with a functioning brain will be possible, allowing re-use and reconfiguration of robot modules. Representations and algorithms will be designed, developed, and experimentally assessed following the common best practices of the research fields that are relevant to the project.

  • OsloMet, Department of Computer Science
  • University of Trieste, Department of Engineering and Architecture
  • Østfold University College, Department of Computer Science and Communication