Engineered self-organization for resilient robot self-assembly with minimal surprise

Abstract

In collective robotic systems, the automatic generation of controllers for complex tasks is still a challenging problem. Open-ended evolution of complex robot behaviors can be a possible solution whereby an intrinsic driver for pattern formation and self-organization may prove to be important. We implement such a driver in collective robot systems by evolving prediction networks as world models in pair with action-selection networks. Fitness is given for good predictions which causes a bias towards easily predictable environments and behaviors in the form of emergent patterns, that is, environments of minimal surprise. There is no task-dependent bias or any other explicit predetermination for the different qualities of the emerging patterns. A careful configuration of actions, sensor models, and the environment is required to stimulate the emergence of complex behaviors. We study self-assembly to increase the scenario's complexity for our minimal surprise approach and, at the same time, limit the complexity of our simulations to a grid world to manage the feasibility of this approach. We investigate the impact of different swarm densities and the shape of the environment on the emergent patterns. Furthermore, we study how evolution can be biased towards the emergence of desired patterns. We analyze the resilience of the resulting self-assembly behaviors by causing damages to the assembled pattern and observe the self-organized reassembly of the structure. In summary, we evolved swarm behaviors for resilient self-assembly and successfully engineered self-organization in simulation. In future work, we plan to transfer our approach to a swarm of real robots.

Evolution of Diverse Swarm Behaviors with Minimal Surprise

“…In all experiments, our swarm of robots lives on a square 2D grid with toroidal boundary conditions (torus). We vary swarm density (i.e., robots per area: N G×G ) by keeping the swarm size N = 100 fixed and changing the grid side lengths G as in previous work (Kaiser and Hamann, 2019). This leads to swarm densities of 44% ( 100 15×15 ) and 25% ( 100 20×20 ), respectively.…”

“…Minimize Surprise Our approach of minimal surprise (Hamann, 2014;Kaiser and Hamann, 2019) allows the evolution of robot (swarm) behaviors solely using an intrinsic driver, here to reach high sensor value prediction accuracy. Each robot in the swarm is equipped with two ANN (see Fig.…”

“…These predefined predictions serve as a local template for the behavior, for example, defining a targeted structure in self-assembly that is similar to approaches used in cellular automata (Hoffmann and Désérable, 2016). We exemplified minimize surprise with predefined predictions aiming for the emergence of line structures in a robot swarm self-assembly approach (Kaiser and Hamann, 2019).…”

“…While this way the approach is task-agnostic and generic, the evolutionary dynamics are biased towards 'boring behaviors,' that is, easy to predict collective behaviors. As it turns out, these are then natural swarm behaviors (Hamann, 2014) and useful self-assembly behaviors (Kaiser and Hamann, 2019). The application of the minimize surprise approach is hence similar to novelty search.…”

“…We address challenge A by testing two behavioral distance measures for novelty search. We address challenge B by using classification scripts that are inspired by experience about the complete set of behaviors emerging in this scenario (Kaiser and Hamann, 2019).…”