Search for: All records

One promising approach towards effective robot decision making in complex, long-horizon tasks is to sequence together parameterized skills. We consider a setting where a robot is initially equipped with (1) a library of parameterized skills, (2) an AI planner for sequencing together the skills given a goal, and (3) a very general prior distribution for selecting skill parameters. Once deployed, the robot should rapidly and autonomously learn to improve its performance by specializing its skill parameter selection policy to the particular objects, goals, and constraints in its environment. In this work, we focus on the active learning problem of choosing which skills to practice to maximize expected future task success. We propose that the robot should estimate the competence of each skill, extrapolate the competence (asking: “how much would the competence improve through practice?”), and situate the skill in the task distribution through competence- aware planning. This approach is implemented within a fully autonomous system where the robot repeatedly plans, practices, and learns without any environment resets. Through experiments in simulation, we find that our approach learns effective pa- rameter policies more sample-efficiently than several baselines. Experiments in the real-world demonstrate our approach’s ability to handle noise from perception and control and improve the robot’s ability to solve two long-horizon mobile-manipulation tasks after a few hours of autonomous practice. Project website: http://ees.csail.mit.edu 

Abstract Background Climate change is expected to lead to warming in ocean surface temperatures which will have unequal effects on the rates of photosynthesis and heterotrophy. As a result of this changing metabolic landscape, directional phenotypic evolution will occur, with implications that cascade up to the ecosystem level. While mixotrophic phytoplankton, organisms that combine photosynthesis and heterotrophy to meet their energetic and nutritional needs, are expected to become more heterotrophic with warmer temperatures due to heterotrophy increasing at a faster rate than photosynthesis, it is unclear how evolution will influence how these organisms respond to warmer temperatures. In this study, we used adaptive dynamics to model the consequences of temperature-mediated increases in metabolic rates for the evolution of mixotrophic phytoplankton, focusing specifically on phagotrophic mixotrophs. Results We find that mixotrophs tend to evolve to become more reliant on phagotrophy as temperatures rise, leading to reduced prey abundance through higher grazing rates. However, if prey abundance becomes too low, evolution favors greater reliance on photosynthesis. These responses depend upon the trade-off that mixotrophs experience between investing in photosynthesis and phagotrophy. Mixotrophs with a convex trade-off maintain mixotrophy over the greatest range of temperatures; evolution in these “generalist” mixotrophs was found to exacerbate carbon cycle impacts, with evolving mixotrophs exhibiting increased sensitivity to rising temperature. Conclusions Our results show that mixotrophs may respond more strongly to climate change than predicted by phenotypic plasticity alone due to evolutionary shifts in metabolic investment. However, the type of metabolic trade-off experienced by mixotrophs as well as ecological feedback on prey abundance may ultimately limit the extent of evolutionary change along the heterotrophy-phototrophy spectrum. 

Abstract Different modes of non‐genetic inheritance are expected to affect population persistence in fluctuating environments. We here analyseCaenorhabditis elegansdensity‐independent per capita growth rate time series on 36 populations experiencing six controlled sequences of challenging oxygen level fluctuations across 60 generations, and parameterise competing models of non‐genetic inheritance in order to explain observed dynamics. Our analysis shows that phenotypic plasticity and anticipatory maternal effects are sufficient to explain growth rate dynamics, but that a carryover model where ‘epigenetic’ memory is imperfectly transmitted and might be reset at each generation is a better fit to the data. We further find that this epigenetic memory is asymmetric since it is kept for longer when populations are exposed to the more challenging environment. Our analysis suggests that population persistence in fluctuating environments depends on the non‐genetic inheritance of phenotypes whose expression is regulated across multiple generations.