Source-linked AI summary
Learning Efficient Navigation in Vortical Flow Fields
Peter Gunnarson, Ioannis Mandralis, Guido Novati, Petros Koumoutsakos, John O. Dabiri
TL;DR
Robots navigating unsteady flows need to choose which local flow cues are useful, since those cues vary and may be limited to immediate surroundings. The paper uses deep reinforcement learning to steer a swimmer through a two-dimensional unsteady flow field, finding near-100% success with local velocity measurements versus 47.2% with vorticity sensing.
Problem
The paper asks which environmental cues are most useful for RL navigation through flow fields, an important challenge for robots operating in unsteady currents.
Source: p. 1Method
Deep reinforcement learning learns a swimmer’s direction of motion from its state, with local flow information supplied as a cue.
Source: p. 2Results
The velocity swimmer achieved near 100% success, compared with a 47.2% success rate for the vorticity RL swimmer.
Source: p. 4Takeaways & Limitations
The findings suggest RL coupled with an onboard velocity sensor may be an effective tool for robot navigation.
Source: p. 6Real robots face additional complexity not considered in the simplified simulation, making their next position harder to determine from local velocity measurements.
Source: p. 4
Abstract
from arXiv · showhide
Efficient point-to-point navigation in the presence of a background flow field is important for robotic applications such as ocean surveying. In such applications, robots may only have knowledge of their immediate surroundings or be faced with time-varying currents, which limits the use of optimal control techniques for planning trajectories. Here, we apply a novel Reinforcement Learning algorithm to discover time-efficient navigation policies to steer a fixed-speed swimmer through an unsteady two-dimensional flow field. The algorithm entails inputting environmental cues into a deep neural network that determines the swimmer's actions, and deploying Remember and Forget Experience replay. We find that the resulting swimmers successfully exploit the background flow to reach the target, but that this success depends on the type of sensed environmental cue. Surprisingly, a velocity sensing approach outperformed a bio-mimetic vorticity sensing approach by nearly two-fold in success rate. Equipped with local velocity measurements, the reinforcement learning algorithm achieved near 100% success in reaching the target locations while approaching the time-efficiency of paths found by a global optimal control planner.