In September 2016, Stanford's "One Hundred Year Study on Artificial Intelligence" project (AI100) issued the first report of its planned long-term periodic assessment of artificial intelligence (AI) and its impact on society. It was written by a panel of 17 study authors, each of whom is deeply rooted in AI research, chaired by Peter Stone of the University of Texas at Austin. The report, entitled "Artificial Intelligence and Life in 2030," examines eight domains of typical urban settings on which AI is likely to have impact over the coming years: transportation, home and service robots, healthcare, education, public safety and security, low-resource communities, employment and workplace, and entertainment. It aims to provide the general public with a scientifically and technologically accurate portrayal of the current state of AI and its potential and to help guide decisions in industry and governments, as well as to inform research and development in the field. The charge for this report was given to the panel by the AI100 Standing Committee, chaired by Barbara Grosz of Harvard University.

In recent years, reinforcement learning (RL) has shown great potential for solving tasks in well-defined environments like games or robotics. This paper aims to solve the robotic reaching task in a simulation run on the Neurorobotics Platform (NRP). The target position is initialized randomly and the robot has 6 degrees of freedom. We compare the performance of various state-of-the-art model-free algorithms. At first, the agent is trained on ground truth data from the simulation to reach the target position in only one continuous movement. Later the complexity of the task is increased by using image data as input from the simulation environment. Experimental results show that training efficiency and results can be improved with appropriate dynamic training schedule function for curriculum learning.

Reinforcement learning (rl) is a popular paradigm for sequential decision making problems. The past decade's advances in rl have led to breakthroughs in many challenging domains such as video games, board games, robotics, and chip design. The sample inefficiency of deep reinforcement learning methods is a significant obstacle when applying rl to real-world problems. Transfer learning has been applied to reinforcement learning such that the knowledge gained in one task can be applied when training in a new task. Curriculum learning is concerned with sequencing tasks or data samples such that knowledge can be transferred between those tasks to learn a target task that would otherwise be too difficult to solve. Designing a curriculum that improves sample efficiency is a complex problem. In this thesis, we propose a teacher-student curriculum learning setting where we simultaneously train a teacher that selects tasks for the student while the student learns how to solve the selected task. Our method is independent of human domain knowledge and manual curriculum design. We evaluated our methods on two reinforcement learning benchmarks: grid world and the challenging Google Football environment. With our method, we can improve the sample efficiency and generality of the student compared to tabula-rasa reinforcement learning.

Card game AI has always been a hot topic in the research of artificial intelligence. In recent years, complex card games such as Mahjong, DouDizhu and Texas Hold'em have been solved and the corresponding AI programs have reached the level of human experts. In this paper, we are devoted to developing an AI program for a more complex card game, GuanDan, whose rules are similar to DouDizhu but much more complicated. To be specific, the characteristics of large state and action space, long length of one episode and the unsure number of players in the GuanDan pose great challenges for the development of the AI program. To address these issues, we propose the first AI program DanZero for GuanDan using reinforcement learning technique. Specifically, we utilize a distributed framework to train our AI system. In the actor processes, we carefully design the state features and agents generate samples by self-play. In the learner process, the model is updated by Deep Monte-Carlo Method. After training for 30 days using 160 CPUs and 1 GPU, we get our DanZero bot. We compare it with 8 baseline AI programs which are based on heuristic rules and the results reveal the outstanding performance of DanZero. We also test DanZero with human players and demonstrate its human-level performance.

Cooperative multi-agent reinforcement learning (MARL) where a team of agents learn coordinated policies optimizing global team rewards has been extensively studied in recent years [25, 13], and find potential applications in a wide variety of domains like robot swarm control [15, 2], coordinating autonomous drivers [26, 41], network routing [38, 4], etc. Although cooperative MARL problems can be framed as a centralized single-agent, with the team as that actor with the joint action space, such an approach doesn't scale well. Joint action space grows exponentially with number of agents in such scenarios. Moreover, due to real world constraints on communication and observability, such framing is often not useful for a large number of real world applications. Unfortunately, simply independently learning decentralized policies based on local observations result into unstable learning and convergence issues due to non-stationarity from simultaneous exploration [12, 33]. This has resulted in MARL methods focusing on the centralized training decentralized execution (CTDE) paradigm, where during training decentralized polices can have access to extra state information during training but not during evaluation.

Abstract-- In robotic manipulation, acquiring samples is extremely expensive because it often requires interacting with the real world. Traditional image-level data augmentation has shown the potential to improve sample efficiency in various machine learning tasks. However, image-level data augmentation is insufficient for an imitation learning agent to learn good manipulation policies in a reasonable amount of demonstrations. We propose Simulation-augmented Equivariant Imitation Learning (SEIL), a method that combines a novel data augmentation strategy of supplementing expert trajectories with simulated transitions and an equivariant model that exploits the O(2) symmetry in robotic manipulation. Experimental evaluations demonstrate that our method can learn non-trivial manipulation tasks within ten demonstrations and outperforms the baselines with a significant margin.

Inverse reinforcement learning (IRL) aims to recover the reward function and the associated optimal policy that best fits observed sequences of states and actions implemented by an expert. Many algorithms for IRL have an inherently nested structure: the inner loop finds the optimal policy given parametrized rewards while the outer loop updates the estimates towards optimizing a measure of fit. For high dimensional environments such nested-loop structure entails a significant computational burden. To reduce the computational burden of a nested loop, novel methods such as SQIL [1] and IQ-Learn [2] emphasize policy estimation at the expense of reward estimation accuracy. However, without accurate estimated rewards, it is not possible to do counterfactual analysis such as predicting the optimal policy under different environment dynamics and/or learning new tasks. In this paper we develop a novel single-loop algorithm for IRL that does not compromise reward estimation accuracy. In the proposed algorithm, each policy improvement step is followed by a stochastic gradient step for likelihood maximization. We show that the proposed algorithm provably converges to a stationary solution with a finite-time guarantee. If the reward is parameterized linearly, we show the identified solution corresponds to the solution of the maximum entropy IRL problem. Finally, by using robotics control problems in MuJoCo and their transfer settings, we show that the proposed algorithm achieves superior performance compared with other IRL and imitation learning benchmarks.

Goal-conditioned reinforcement learning (RL) is a promising direction for training agents that are capable of solving multiple tasks and reach a diverse set of objectives. How to \textit{specify} and \textit{ground} these goals in such a way that we can both reliably reach goals during training as well as generalize to new goals during evaluation remains an open area of research. Defining goals in the space of noisy and high-dimensional sensory inputs poses a challenge for training goal-conditioned agents, or even for generalization to novel goals. We propose to address this by learning factorial representations of goals and processing the resulting representation via a discretization bottleneck, for coarser goal specification, through an approach we call DGRL. We show that applying a discretizing bottleneck can improve performance in goal-conditioned RL setups, by experimentally evaluating this method on tasks ranging from maze environments to complex robotic navigation and manipulation. Additionally, we prove a theorem lower-bounding the expected return on out-of-distribution goals, while still allowing for specifying goals with expressive combinatorial structure.

In a connected transportation system, adaptive traffic signal controllers (ATSC) utilize real-time vehicle trajectory data received from vehicles through wireless connectivity (i.e., connected vehicles) to regulate green time. However, this wirelessly connected ATSC increases cyber-attack surfaces and increases their vulnerability to various cyber-attack modes, which can be leveraged to induce significant congestion in a roadway network. An attacker may receive financial benefits to create such a congestion for a specific roadway. One such mode is a 'sybil' attack in which an attacker creates fake vehicles in the network by generating fake Basic Safety Messages (BSMs) imitating actual connected vehicles following roadway traffic rules. The ultimate goal of an attacker will be to block a route(s) by generating fake or 'sybil' vehicles at a rate such that the signal timing and phasing changes occur without flagging any abrupt change in number of vehicles. Because of the highly non-linear and unpredictable nature of vehicle arrival rates and the ATSC algorithm, it is difficult to find an optimal rate of sybil vehicles, which will be injected from different approaches of an intersection. Thus, it is necessary to develop an intelligent cyber-attack model to prove the existence of such attacks. In this study, a reinforcement learning based cyber-attack model is developed for a waiting time-based ATSC. Specifically, an RL agent is trained to learn an optimal rate of sybil vehicle injection to create congestion for an approach(s). Our analyses revealed that the RL agent can learn an optimal policy for creating an intelligent attack.

Deep reinforcement learning (RL) based controllers for legged robots have demonstrated impressive robustness for walking in different environments for several robot platforms. To enable the application of RL policies for humanoid robots in real-world settings, it is crucial to build a system that can achieve robust walking in any direction, on 2D and 3D terrains, and be controllable by a user-command. In this paper, we tackle this problem by learning a policy to follow a given step sequence. The policy is trained with the help of a set of procedurally generated step sequences (also called footstep plans). We show that simply feeding the upcoming 2 steps to the policy is sufficient to achieve omnidirectional walking, turning in place, standing, and climbing stairs. Our method employs curriculum learning on the complexity of terrains, and circumvents the need for reference motions or pre-trained weights. We demonstrate the application of our proposed method to learn RL policies for 2 new robot platforms - HRP5P and JVRC-1 - in the MuJoCo simulation environment. The code for training and evaluation is available online.