We propose firefly neural architecture descent, a general framework for progressively and dynamically growing neural networks to jointly optimize the networks' parameters and architectures. Our method works in a steepest descent fashion, which iteratively finds the best network within a functional neighborhood of the original network that includes a diverse set of candidate network structures. By using Taylor approximation, the optimal network structure in the neighborhood can be found with a greedy selection procedure. We show that firefly descent can flexibly grow networks both wider and deeper, and can be applied to learn accurate but resource-efficient neural architectures that avoid catastrophic forgetting in continual learning. Empirically, firefly descent achieves promising results on both neural architecture search and continual learning. In particular, on a challenging continual image classification task, it learns networks that are smaller in size but have higher average accuracy than those learned by the state-of-the-art methods.
Moving in complex environments is an essential capability of intelligent mobile robots. Decades of research and engineering have been dedicated to developing sophisticated navigation systems to move mobile robots from one point to another. Despite their overall success, a recently emerging research thrust is devoted to developing machine learning techniques to address the same problem, based in large part on the success of deep learning. However, to date, there has not been much direct comparison between the classical and emerging paradigms to this problem. In this article, we survey recent works that apply machine learning for motion control in mobile robot navigation, within the context of classical navigation systems. The surveyed works are classified into different categories, which delineate the relationship of the learning approaches to classical methods. Based on this classification, we identify common challenges and promising future directions.
While classical autonomous navigation systems can typically move robots from one point to another safely and in a collision-free manner, these systems may fail or produce suboptimal behavior in certain scenarios. The current practice in such scenarios is to manually re-tune the system's parameters, e.g. max speed, sampling rate, inflation radius, to optimize performance. This practice requires expert knowledge and may jeopardize performance in the originally good scenarios. Meanwhile, it is relatively easy for a human to identify those failure or suboptimal cases and provide a teleoperated intervention to correct the failure or suboptimal behavior. In this work, we seek to learn from those human interventions to improve navigation performance. In particular, we propose Adaptive Planner Parameter Learning from Interventions (APPLI), in which multiple sets of navigation parameters are learned during training and applied based on a confidence measure to the underlying navigation system during deployment. In our physical experiments, the robot achieves better performance compared to the planner with static default parameters, and even dynamic parameters learned from a full human demonstration. We also show APPLI's generalizability in another unseen physical test course and a suite of 300 simulated navigation environments.
Classical navigation systems typically operate using a fixed set of hand-picked parameters (e.g. maximum speed, sampling rate, inflation radius, etc.) and require heavy expert re-tuning in order to work in new environments. To mitigate this requirement, it has been proposed to learn parameters for different contexts in a new environment using human demonstrations collected via teleoperation. However, learning from human demonstration limits deployment to the training environment, and limits overall performance to that of a potentially-suboptimal demonstrator. In this paper, we introduce APPLR, Adaptive Planner Parameter Learning from Reinforcement, which allows existing navigation systems to adapt to new scenarios by using a parameter selection scheme discovered via reinforcement learning (RL) in a wide variety of simulation environments. We evaluate APPLR on a robot in both simulated and physical experiments, and show that it can outperform both a fixed set of hand-tuned parameters and also a dynamic parameter tuning scheme learned from human demonstration.
Deep reinforcement learning (RL) algorithms are powerful tools for solving visuomotor decision tasks. However, the trained models are often difficult to interpret, because they are represented as end-to-end deep neural networks. In this paper, we shed light on the inner workings of such trained models by analyzing the pixels that they attend to during task execution, and comparing them with the pixels attended to by humans executing the same tasks. To this end, we investigate the following two questions that, to the best of our knowledge, have not been previously studied. 1) How similar are the visual features learned by RL agents and humans when performing the same task? and, 2) How do similarities and differences in these learned features correlate with RL agents' performance on these tasks? Specifically, we compare the saliency maps of RL agents against visual attention models of human experts when learning to play Atari games. Further, we analyze how hyperparameters of the deep RL algorithm affect the learned features and saliency maps of the trained agents. The insights provided by our results have the potential to inform novel algorithms for the purpose of closing the performance gap between human experts and deep RL agents.
The year 2020 has seen the COVID-19 virus lead to one of the worst global pandemics in history. As a result, governments around the world are faced with the challenge of protecting public health, while keeping the economy running to the greatest extent possible. Epidemiological models provide insight into the spread of these types of diseases and predict the effects of possible intervention policies. However, to date,the even the most data-driven intervention policies rely on heuristics. In this paper, we study how reinforcement learning (RL) can be used to optimize mitigation policies that minimize the economic impact without overwhelming the hospital capacity. Our main contributions are (1) a novel agent-based pandemic simulator which, unlike traditional models, is able to model fine-grained interactions among people at specific locations in a community; and (2) an RL-based methodology for optimizing fine-grained mitigation policies within this simulator. Our results validate both the overall simulator behavior and the learned policies under realistic conditions.
Learning motion planners to move robot from one point to another within an obstacle-occupied space in a collision-free manner requires either an extensive amount of data or high-quality demonstrations. This requirement is caused by the fact that among the variety of maneuvers the robot can perform, it is difficult to find the single optimal plan without many trial-and-error or an expert who is already capable of doing so. However, given a plan performed in obstacle-free space, it is relatively easy to find an obstacle geometry, where this plan is optimal. We consider this "dual" problem of classical motion planning and name this process of finding appropriate obstacle geometry as hallucination. In this work, we present two different approaches to hallucinate (1) the most constrained and (2) a minimal obstacle space where a given plan executed during an exploration phase in a completely safe obstacle-free environment remains optimal. We then train an end-to-end motion planner that can produce motions to move through realistic obstacles during deployment. Both methods are tested on a physical mobile robot in real-world cluttered environments.
Learning from Hallucination (LfH) is a recent machine learning paradigm for autonomous navigation, which uses training data collected in completely safe environments and adds numerous imaginary obstacles to make the environment densely constrained, to learn navigation planners that produce feasible navigation even in highly constrained (more dangerous) spaces. However, LfH requires hallucinating the robot perception during deployment to match with the hallucinated training data, which creates a need for sometimes-infeasible prior knowledge and tends to generate very conservative planning. In this work, we propose a new LfH paradigm that does not require runtime hallucination---a feature we call "sober deployment"---and can therefore adapt to more realistic navigation scenarios. This novel Hallucinated Learning and Sober Deployment (HLSD) paradigm is tested in a benchmark testbed of 300 simulated navigation environments with a wide range of difficulty levels, and in the real-world. In most cases, HLSD outperforms both the original LfH method and a classical navigation planner.
Experience replay (ER) improves the data efficiency of off-policy reinforcement learning (RL) algorithms by allowing an agent to store and reuse its past experiences in a replay buffer. While many techniques have been proposed to enhance ER by biasing how experiences are sampled from the buffer, thus far they have not considered strategies for refreshing experiences inside the buffer. In this work, we introduce Lucid Dreaming for Experience Replay (LiDER), a conceptually new framework that allows replay experiences to be refreshed by leveraging the agent's current policy. LiDER 1) moves an agent back to a past state; 2) lets the agent try following its current policy to execute different actions---as if the agent were "dreaming" about the past, but is aware of the situation and can control the dream to encounter new experiences; and 3) stores and reuses the new experience if it turned out better than what the agent previously experienced, i.e., to refresh its memories. LiDER is designed to be easily incorporated into off-policy, multi-worker RL algorithms that use ER; we present in this work a case study of applying LiDER to an actor-critic based algorithm. Results show LiDER consistently improves performance over the baseline in four Atari 2600 games. Our open-source implementation of LiDER and the data used to generate all plots in this paper are available at github.com/duyunshu/lucid-dreaming-for-exp-replay.
Reactions such as gestures, facial expressions, and vocalizations are an abundant, naturally occurring channel of information that humans provide during interactions. A robot or other agent could leverage an understanding of such implicit human feedback to improve its task performance at no cost to the human. This approach contrasts with common agent teaching methods based on demonstrations, critiques, or other guidance that need to be attentively and intentionally provided. In this paper, we first define the general problem of learning from implicit human feedback and then propose to address this problem through a novel data-driven framework, EMPATHIC. This two-stage method consists of (1) mapping implicit human feedback to relevant task statistics such as rewards, optimality, and advantage; and (2) using such a mapping to learn a task. We instantiate the first stage and three second-stage evaluations of the learned mapping. To do so, we collect a dataset of human facial reactions while participants observe an agent execute a sub-optimal policy for a prescribed training task. We train a deep neural network on this data and demonstrate its ability to (1) infer relative reward ranking of events in the training task from prerecorded human facial reactions; (2) improve the policy of an agent in the training task using live human facial reactions; and (3) transfer to a novel domain in which it evaluates robot manipulation trajectories.