Abstract: In some evolutionary robotics experiments, evolved robots are transferred from simulation to reality, while sensor/motor data flows back from reality to improve the next transferral. We envision a generalization of this approach: a simulation-to-reality pipeline. In this pipeline, increasingly embodied agents flow up through a sequence of increasingly physically realistic simulators, while data flows back down to improve the next transferral between neighboring simulators; physical reality is the last link in this chain. As a first proof of concept, we introduce a two-link chain: A fast yet low-fidelity (lo-fi) simulator hosts minimally embodied agents, which gradually evolve controllers and morphologies to colonize a slow yet high-fidelity (hi-fi) simulator. The agents are thus physically scaffolded. We show here that, given the same computational budget, these physically scaffolded robots reach higher performance in the hi-fi simulator than do robots that only evolve in the hi-fi simulator, but only for a sufficiently difficult task. These results suggest that a simulation-to-reality pipeline may strike a good balance between accelerating evolution in simulation while anchoring the results in reality, free the investigator from having to prespecify the robot's morphology, and pave the way to scalable, automated, robot-generating systems.
Abstract: Infrastructure for the automatic collection of single-point measurements of snow water equivalent (SWE) is well-established. However, because SWE varies significantly over space, the estimation of SWE at the catchment scale based on a single-point measurement is error-prone. We propose low-cost, lightweight methods for near-real-time estimation of mean catchment-wide SWE using existing infrastructure, wireless sensor networks, and machine learning algorithms. Because snowpack distribution is highly nonlinear, we focus on Genetic Programming (GP), a nonlinear, white-box, inductive machine learning algorithm. Because we did not have access to near-real-time catchment-scale SWE data, we used available data as ground truth for machine learning in a set of experiments that are successive approximations of our goal of catchment-wide SWE estimation. First, we used a history of maritime snowpack data collected by manual snow courses. Second, we used distributed snow depth (HS) data collected automatically by wireless sensor networks. We compared the performance of GP against linear regression (LR), binary regression trees (BT), and a widely used basic method (BM) that naively assumes non-variable snowpack. In the first experiment set, GP and LR models predicted SWE with lower error than BM. In the second experiment set, GP had lower error than LR, but outperformed BT only when we applied a technique that specifically mitigated the possibility of over-fitting.