So, what is reinforcement learning? It's a training technique that teaches AI complex behavior by trial and error. In other words, the robot tries to walk and learns from its mistakes whenever it fails, similar to how babies learn how to walk. But before Cassie could stumble around in the real world, the team started with a simulation of the robot in a virtual world. Using a training environment called MuJoCo, the simulated robot referenced a large library of possible movements and learned how to apply them. The team then ran a second simulation, called Matlab SimMechanics, to test the robot in simulated real-world conditions.

Autonomous driving has emerged as one of the most active areas of research as it has the promise of making transportation safer and more efficient than ever before. Most real-world autonomous driving pipelines perform perception, motion planning and action in a loop. In this work we present MADRaS, an open-source multi-agent driving simulator for use in the design and evaluation of motion planning algorithms for autonomous driving. Given a start and a goal state, the task of motion planning is to solve for a sequence of position, orientation and speed values in order to navigate between the states while adhering to safety constraints. These constraints often involve the behaviors of other agents in the environment. MADRaS provides a platform for constructing a wide variety of highway and track driving scenarios where multiple driving agents can be trained for motion planning tasks using reinforcement learning and other machine learning algorithms. MADRaS is built on TORCS, an open-source car-racing simulator. TORCS offers a variety of cars with different dynamic properties and driving tracks with different geometries and surface.  MADRaS inherits these functionalities from TORCS and introduces support for multi-agent training, inter-vehicular communication, noisy observations, stochastic actions, and custom traffic cars whose behaviors can be programmed to simulate challenging traffic conditions encountered in the real world. MADRaS can be used to create driving tasks whose complexities can be tuned along eight axes in well-defined steps. This makes it particularly suited for curriculum and continual learning. MADRaS is lightweight and it provides a convenient OpenAI Gym interface for independent control of each car. Apart from the primitive steering-acceleration-brake control mode of TORCS, MADRaS offers a hierarchical track-position – speed control mode that can potentially be used to achieve better generalization. MADRaS uses a UDP based client server model where the simulation engine is the server and each client is a driving agent. MADRaS uses multiprocessing to run each agent as a parallel process for efficiency and integrates well with popular reinforcement learning libraries like RLLib. We show experiments on single and multi-agent reinforcement learning with and without curriculum

Artificial Intelligence (AI) can augment and sometimes even replace human cognition. Inspired by efforts to value human agency alongside productivity, we discuss the benefits of solving Citizen Science (CS) tasks with Hybrid Intelligence (HI), a synergetic mixture of human and artificial intelligence. Currently there is no clear framework or methodology on how to create such an effective mixture. Due to the unique participant-centered set of values and the abundance of tasks drawing upon both human common sense and complex 21st century skills, we believe that the field of CS offers an invaluable testbed for the development of HI and human-centered AI of the 21st century, while benefiting CS as well. In order to investigate this potential, we first relate CS to adjacent computational disciplines. Then, we demonstrate that CS projects can be grouped according to their potential for HI-enhancement by examining two key dimensions: the level of digitization and the amount of knowledge or experience required for participation. Finally, we propose a framework for types of human-AI interaction in CS based on established criteria of HI. This "HI lens" provides the CS community with an overview of several ways to utilize the combination of AI and human intelligence in their projects. It also allows the AI community to gain ideas on how developing AI in CS projects can further their own field.

Collision avoidance algorithms are of central interest to many drone applications. In particular, decentralized approaches may be the key to enabling robust drone swarm solutions in cases where centralized communication becomes computationally prohibitive. In this work, we draw biological inspiration from flocks of starlings (Sturnus vulgaris) and apply the insight to end-to-end learned decentralized collision avoidance. More specifically, we propose a new, scalable observation model following a biomimetic topological interaction rule that leads to stable learning and robust avoidance behavior. Additionally, prior work primarily focuses on invoking a separation principle, i.e. designing collision avoidance independent of specific tasks. By applying a general reinforcement learning approach, we propose a holistic learning-based approach to integrating collision avoidance with various tasks and dynamics. To validate the generality of this approach, we successfully apply our methodology to a number of configurations. Our learned policies are tested in simulation and subsequently transferred to real-world drones to validate their real-world applicability.

As Deep Learning continues to drive a variety of applications in datacenters and HPC, there is a growing trend towards building large accelerators with several sub-accelerator cores/chiplets. This work looks at the problem of supporting multi-tenancy on such accelerators. In particular, we focus on the problem of mapping layers from several DNNs simultaneously on an accelerator. Given the extremely large search space, we formulate the search as an optimization problem and develop a specialized genetic algorithm called G# withcustom operators to enable structured sample-efficient exploration. We quantitatively compare G# with several common heuristics, state-of-the-art optimization methods, and reinforcement learning methods across different accelerator set-tings (large/small accelerators) and different sub-accelerator configurations (homogeneous/heterogeneous), and observeG# can consistently find better solutions. Further, to enable real-time scheduling, we also demonstrate a method to generalize the learnt schedules and transfer them to the next batch of jobs, reducing schedule compute time to near zero.

Recently, Intelligent Transportation Systems are leveraging the power of increased sensory coverage and computing power to deliver data-intensive solutions achieving higher levels of performance than traditional systems. Within Traffic Signal Control (TSC), this has allowed the emergence of Machine Learning (ML) based systems. Among this group, Reinforcement Learning (RL) approaches have performed particularly well. Given the lack of industry standards in ML for TSC, literature exploring RL often lacks comparison against commercially available systems and straightforward formulations of how the agents operate. Here we attempt to bridge that gap. We propose three different architectures for TSC RL agents and compare them against the currently used commercial systems MOVA, SurTrac and Cyclic controllers and provide pseudo-code for them. The agents use variations of Deep Q-Learning and Actor Critic, using states and rewards based on queue lengths. Their performance is compared in across different map scenarios with variable demand, assessing them in terms of the global delay and average queue length. We find that the RL-based systems can significantly and consistently achieve lower delays when compared with existing commercial systems.

As many of you know, I am doing my PhD centered around model-based reinforcement learning (MBRL). This post is not talking about the technical details and recent work, but rather why I am bullish on it for the future. Beyond the prospects of how well it can perform (it's much younger than most of deep RL), having discussions with AI Safety and Ethical AI experts makes it clear that it's more structured learning-setup is pointing towards systems that humans can better understand. Some level of understanding how the system makes decisions is likely a prerequisite for many companies to start using it, else they cannot do real A/B testing and analysis. I will start by showing you the rich set of parallels MBRL has in biological processes, and then show the features making it more suitable for safe deployment in society-facing systems (see why this matters here).

Recurrent meta reinforcement learning (meta-RL) agents are agents that employ a recurrent neural network (RNN) for the purpose of "learning a learning algorithm". After being trained on a pre-specified task distribution, the learned weights of the agent's RNN are said to implement an efficient learning algorithm through their activity dynamics, which allows the agent to quickly solve new tasks sampled from the same distribution. However, due to the black-box nature of these agents, the way in which they work is not yet fully understood. In this study, we shed light on the internal working mechanisms of these agents by reformulating the meta-RL problem using the Partially Observable Markov Decision Process (POMDP) framework. We hypothesize that the learned activity dynamics is acting as belief states for such agents. Several illustrative experiments suggest that this hypothesis is true, and that recurrent meta-RL agents can be viewed as agents that learn to act optimally in partially observable environments consisting of multiple related tasks. This view helps in understanding their failure cases and some interesting model-based results reported in the literature.

Balancing and push-recovery are essential capabilities enabling humanoid robots to solve complex locomotion tasks. In this context, classical control systems tend to be based on simplified physical models and hard-coded strategies. Although successful in specific scenarios, this approach requires demanding tuning of parameters and switching logic between specifically-designed controllers for handling more general perturbations. We apply model-free Deep Reinforcement Learning for training a general and robust humanoid push-recovery policy in a simulation environment. Our method targets high-dimensional whole-body humanoid control and is validated on the iCub humanoid. Reward components incorporating expert knowledge on humanoid control enable fast learning of several robust behaviors by the same policy, spanning the entire body. We validate our method with extensive quantitative analyses in simulation, including out-of-sample tasks which demonstrate policy robustness and generalization, both key requirements towards real-world robot deployment.

Visual domain randomization in simulated environments is a widely used method to transfer policies trained in simulation to real robots. However, domain randomization and augmentation hamper the training of a policy. As reinforcement learning struggles with a noisy training signal, this additional nuisance can drastically impede training. For difficult tasks it can even result in complete failure to learn. To overcome this problem we propose to pre-train a perception encoder that already provides an embedding invariant to the randomization. We demonstrate that this yields consistently improved results on a randomized version of DeepMind control suite tasks and a stacking environment on arbitrary backgrounds with zero-shot transfer to a physical robot.