Abstract--Grasping small objects surrounded by unstable or non-rigid material plays a crucial role in applications such as surgery, harvesting, construction, disaster recovery, and assisted feeding. This task is especially difficult when fine manipulation is required in the presence of sensor noise and perception errors; errors inevitably trigger dynamic motion, which is challenging to model precisely. Circumventing the difficulty to build accurate models for contacts and dynamics, data-driven methods like reinforcement learning (RL) can optimize task performance via trial and error, reducing the need for accurate models of contacts and dynamics. Applying RL methods to real robots, however, has been hindered by factors such as prohibitively high sample complexity or the high training infrastructure cost for providing resets on hardware. This work presents CherryBot, an RL system that uses chopsticks for fine manipulation that surpasses human reactiveness for some dynamic grasping tasks. By integrating imprecise simulators, suboptimal demonstrations and external state estimation, we study how to make a realworld robot learning system sample efficient and general while reducing the human effort required for supervision. Our system shows continual improvement through 30 minutes of real-world interaction: through reactive retry, it achieves an almost 100% success rate on the demanding task of using chopsticks to grasp small objects swinging in the air. We demonstrate the reactiveness, robustness and generalizability of CherryBot to varying object shapes and dynamics (e.g., external disturbances However, this research investigates a more universal solution: assuming that fine manipulation is required, inaccuracy is How can we automate the task of picking cherries from a unavoidable and real-time reaction is necessary, can we enable tree branch that is blowing in the wind, causing the branch dynamic fine grasping without stable support? An ideal agent to shake and the cherries to tremble? This scenario is an should be: example of fine grasping without rigid-surface support, and its challenges are two-fold.

Various methods for Multi-Agent Reinforcement Learning (MARL) have been developed with the assumption that agents' policies are based on accurate state information. However, policies learned through Deep Reinforcement Learning (DRL) are susceptible to adversarial state perturbation attacks. In this work, we propose a State-Adversarial Markov Game (SAMG) and make the first attempt to investigate the fundamental properties of MARL under state uncertainties. Our analysis shows that the commonly used solution concepts of optimal agent policy and robust Nash equilibrium do not always exist in SAMGs. To circumvent this difficulty, we consider a new solution concept called robust agent policy, where agents aim to maximize the worst-case expected state value. We prove the existence of robust agent policy for finite state and finite action SAMGs. Additionally, we propose a Robust Multi-Agent Adversarial Actor-Critic (RMA3C) algorithm to learn robust policies for MARL agents under state uncertainties. Our experiments demonstrate that our algorithm outperforms existing methods when faced with state perturbations and greatly improves the robustness of MARL policies. Our code is public on https://songyanghan.github.io/what_is_solution/.

Our collision avoidance system SAFER takes input from Mobile robots are slowly but surely taking a place in our lidar and ultrasonic sensor scans, wheel odometry for robot everyday lives and work environments with various applications: state, and the upstream control commands. We fuse the vacuum cleaning, video recording, companionship, lidar and ultrasonic sensor scans to detect a diverse set security, tele-presence, etc. Whether they are autonomous of obstacles, including transparent glass, reflective surfaces, agents or controlled by human operators, collision avoidance furniture, humans, etc. We design a reward function for is key for mobile agents to operate safely, and effectively our RL agent with two terms. The first term encourages in the real world. There are numerous approaches to collision the reduction of AEB activation. The second term improves avoidance, including search-based planning methods, collision avoidance metrics through a cost function, such as trajectory optimization, learning-based methods, and emergency average speed, distance to obstacles, and matching human intervention systems.

For decades, people have been seeking for fishlike flapping motions that can realize underwater propulsion with low energy cost. Complexity of the nonstationary flow field around the flapping body makes this problem very difficult. In earlier studies, motion patterns are usually prescribed as certain periodic functions which constrains the following optimization process in a small subdomain of the whole motion space. In this work, to avoid this motion constraint, a variational autoencoder (VAE) is designed to compress various flapping motions into a simple action vector. Then we let a flapping airfoil continuously interact with water tunnel environment and adjust its action accordingly through a reinforcement learning (RL) framework. By this automatic close-looped experiment, we obtain several motion patterns that can result in high hydrodynamic efficiency comparing to pure harmonic motions with the same thrust level. And we find that, after numerous trials and errors, RL trainings in current experiment always converge to motion patterns that are close to harmonic motions. In other words, current work proves that harmonic motion with appropriate amplitude and frequency is always an optimal choice for efficient underwater propulsion. Furthermore, the RL framework proposed here can be also extended to the study of other complex swimming problems, which might pave the way for the creation of a robotic fish that can swim like a real fish.

Cybersecurity solutions have shown promising performance when detecting ransomware samples that use fixed algorithms and encryption rates. However, due to the current explosion of Artificial Intelligence (AI), sooner than later, ransomware (and malware in general) will incorporate AI techniques to intelligently and dynamically adapt its encryption behavior to be undetected. It might result in ineffective and obsolete cybersecurity solutions, but the literature lacks AI-powered ransomware to verify it. Thus, this work proposes RansomAI, a Reinforcement Learning-based framework that can be integrated into existing ransomware samples to adapt their encryption behavior and stay stealthy while encrypting files. RansomAI presents an agent that learns the best encryption algorithm, rate, and duration that minimizes its detection (using a reward mechanism and a fingerprinting intelligent detection system) while maximizing its damage function. The proposed framework was validated in a ransomware, Ransomware-PoC, that infected a Raspberry Pi 4, acting as a crowdsensor. A pool of experiments with Deep Q-Learning and Isolation Forest (deployed on the agent and detection system, respectively) has demonstrated that RansomAI evades the detection of Ransomware-PoC affecting the Raspberry Pi 4 in a few minutes with >90% accuracy.

Deep reinforcement learning (RL) has brought many successes for autonomous robot navigation. However, there still exists important limitations that prevent real-world use of RL-based navigation systems. For example, most learning approaches lack safety guarantees; and learned navigation systems may not generalize well to unseen environments. Despite a variety of recent learning techniques to tackle these challenges in general, a lack of an open-source benchmark and reproducible learning methods specifically for autonomous navigation makes it difficult for roboticists to choose what learning methods to use for their mobile robots and for learning researchers to identify current shortcomings of general learning methods for autonomous navigation. In this paper, we identify four major desiderata of applying deep RL approaches for autonomous navigation: (D1) reasoning under uncertainty, (D2) safety, (D3) learning from limited trial-and-error data, and (D4) generalization to diverse and novel environments. Then, we explore four major classes of learning techniques with the purpose of achieving one or more of the four desiderata: memory-based neural network architectures (D1), safe RL (D2), model-based RL (D2, D3), and domain randomization (D4). By deploying these learning techniques in a new open-source large-scale navigation benchmark and real-world environments, we perform a comprehensive study aimed at establishing to what extent can these techniques achieve these desiderata for RL-based navigation systems.

Training AI with strong and rich strategies in multi-agent environments remains an important research topic in Deep Reinforcement Learning (DRL). The AI's strength is closely related to its diversity of strategies, and this relationship can guide us to train AI with both strong and rich strategies. To prove this point, we propose Diversity is Strength (DIS), a novel DRL training framework that can simultaneously train multiple kinds of AIs. These AIs are linked through an interconnected history model pool structure, which enhances their capabilities and strategy diversities. We also design a model evaluation and screening scheme to select the best models to enrich the model pool and obtain the final AI. The proposed training method provides diverse, generalizable, and strong AI strategies without using human data. We tested our method in an AI competition based on Google Research Football (GRF) and won the 5v5 and 11v11 tracks. The method enables a GRF AI to have a high level on both 5v5 and 11v11 tracks for the first time, which are under complex multi-agent environments. The behavior analysis shows that the trained AI has rich strategies, and the ablation experiments proved that the designed modules benefit the training process.

Recent advances in high-fidelity simulators have enabled closed-loop training of autonomous driving agents, potentially solving the distribution shift in training v.s. deployment and allowing training to be scaled both safely and cheaply. However, there is a lack of understanding of how to build effective training benchmarks for closed-loop training. In this work, we present the first empirical study which analyzes the effects of different training benchmark designs on the success of learning agents, such as how to design traffic scenarios and scale training environments. Furthermore, we show that many popular RL algorithms cannot achieve satisfactory performance in the context of autonomous driving, as they lack long-term planning and take an extremely long time to train. To address these issues, we propose trajectory value learning (TRAVL), an RL-based driving agent that performs planning with multistep look-ahead and exploits cheaply generated imagined data for efficient learning. Our experiments show that TRAVL can learn much faster and produce safer maneuvers compared to all the baselines. For more information, visit the project website: https://waabi.ai/research/travl

Many policy-based reinforcement learning (RL) algorithms can be viewed as instantiations of approximate policy iteration (PI), i.e., where policy improvement and policy evaluation are both performed approximately. In applications where the average reward objective is the meaningful performance metric, discounted reward formulations are often used with the discount factor being close to $1,$ which is equivalent to making the expected horizon very large. However, the corresponding theoretical bounds for error performance scale with the square of the horizon. Thus, even after dividing the total reward by the length of the horizon, the corresponding performance bounds for average reward problems go to infinity. Therefore, an open problem has been to obtain meaningful performance bounds for approximate PI and RL algorithms for the average-reward setting. In this paper, we solve this open problem by obtaining the first finite-time error bounds for average-reward MDPs, and show that the asymptotic error goes to zero in the limit as policy evaluation and policy improvement errors go to zero.

Neutral atoms devices represent a promising technology that uses optical tweezers to geometrically arrange atoms and modulated laser pulses to control the quantum states. A neutral atoms Noisy Intermediate Scale Quantum (NISQ) device is developed by Pasqal with rubidium atoms that will allow to work with up to 100 qubits. All NISQ devices are affected by noise that have an impact on the computations results. Therefore it is important to better understand and characterize the noise sources and possibly to correct them. Here, two approaches are proposed to characterize and correct noise parameters on neutral atoms NISQ devices. In particular the focus is on Pasqal devices and Machine Learning (ML) techniques are adopted to pursue those objectives. To characterize the noise parameters, several ML models are trained, using as input only the measurements of the final quantum state of the atoms, to predict laser intensity fluctuation and waist, temperature and false positive and negative measurement rate. Moreover, an analysis is provided with the scaling on the number of atoms in the system and on the number of measurements used as input. Also, we compare on real data the values predicted with ML with the a priori estimated parameters. Finally, a Reinforcement Learning (RL) framework is employed to design a pulse in order to correct the effect of the noise in the measurements. It is expected that the analysis performed in this work will be useful for a better understanding of the quantum dynamic in neutral atoms devices and for the widespread adoption of this class of NISQ devices.