In this paper, we present a deep reinforcement learning method for quadcopter bypassing the obstacle on the flying path. In the past study, the algorithm only controls the forward direction about quadcopter. In this letter, we use two functions to control quadcopter. One is quadcopter navigating function. It is based on calculating coordination point and find the straight path to the goal. The other function is collision avoidance function. It is implemented by deep Q-network model. Both two function will output rotating degree, the agent will combine both output and turn direct. Besides, deep Q-network can also make quadcopter fly up and down to bypass the obstacle and arrive at the goal. Our experimental result shows that the collision rate is 14% after 500 flights. Based on this work, we will train more complex sense and transfer model to the real quadcopter.

In this work we seek for an approach to integrate safety in the learning process that relies on a partly known state-space model of the system and regards the unknown dynamics as an additive bounded disturbance. We introduce a framework for safely learning a control strategy for a given system with an additive disturbance. On the basis of the known part of the model, a safe set in which the system can learn safely, the algorithm can choose optimal actions for pursuing the target set as long as the safety-preserving condition is satisfied. After some learning episodes, the disturbance can be updated based on real-world data. To this end, Gaussian Process regression is conducted on the collected disturbance samples. Since the unstable nature of the law of the real world, for example, change of friction or conductivity with the temperature, we expect to have the more robust solution of optimal control problem. For evaluation of approach described above we choose an inverted pendulum as a benchmark model. The proposed algorithm manages to learn a policy that does not violate the pre-specified safety constraints. Observed performance is improved when it was incorporated exploration set up to make sure that an optimal policy is learned everywhere in the safe set. Finally, we outline some promising directions for future research beyond the scope of this paper.

Direct policy search is one of the most important algorithm of reinforcement learning. However, learning from scratch needs a large amount of experience data and can be easily prone to poor local optima. In addition to that, a partially trained policy tends to perform dangerous action to agent and environment. In order to overcome these challenges, this paper proposed a policy initialization algorithm called Policy Learning based on Completely Behavior Cloning (PLCBC). PLCBC first transforms the Model Predictive Control (MPC) controller into a piecewise affine (PWA) function using multi-parametric programming, and uses a neural network to express this function. By this way, PLCBC can completely clone the MPC controller without any performance loss, and is totally training-free. The experiments show that this initialization strategy can help agent learn at the high reward state region, and converge faster and better.

Most existing approaches to autonomous driving fall into one of two categories: modular pipelines, that build an extensive model of the environment, and imitation learning approaches, that map images directly to control outputs. A recently proposed third paradigm, direct perception, aims to combine the advantages of both by using a neural network to learn appropriate low-dimensional intermediate representations. However, existing direct perception approaches are restricted to simple highway situations, lacking the ability to navigate intersections, stop at traffic lights or respect speed limits. In this work, we propose a direct perception approach which maps video input to intermediate representations suitable for autonomous navigation in complex urban environments given high-level directional inputs. Compared to state-of-the-art reinforcement and conditional imitation learning approaches, we achieve an improvement of up to 68 % in goal-directed navigation on the challenging CARLA simulation benchmark. In addition, our approach is the first to handle traffic lights and speed signs by using image-level labels only, as well as smooth car-following, resulting in a significant reduction of traffic accidents in simulation.

Deep learning models have become state of the art for natural language processing (NLP) tasks, however deploying these models in production system poses significant memory constraints. Existing compression methods are either lossy or introduce significant latency. We propose a compression method that leverages low rank matrix factorization during training,to compress the word embedding layer which represents the size bottleneck for most NLP models. Our models are trained, compressed and then further re-trained on the downstream task to recover accuracy while maintaining the reduced size. Empirically, we show that the proposed method can achieve 90% compression with minimal impact in accuracy for sentence classification tasks, and outperforms alternative methods like fixed-point quantization or offline word embedding compression. We also analyze the inference time and storage space for our method through FLOP calculations, showing that we can compress DNN models by a configurable ratio and regain accuracy loss without introducing additional latency compared to fixed point quantization. Finally, we introduce a novel learning rate schedule, the Cyclically Annealed Learning Rate (CALR), which we empirically demonstrate to outperform other popular adaptive learning rate algorithms on a sentence classification benchmark.

In this paper, we show how behaviour trees (BTs) can be used to design modular, versatile, and robust control architectures for mission-critical systems. In particular, we show this in the context of autonomous underwater vehicles (AUVs). Robustness, in terms of system safety, is important since manual recovery of AUVs is often extremely difficult. Further more, versatility is important to be able to execute many different kinds of missions. Finally, modularity is needed to achieve a combination of robustness and versatility, as the complexity of a versatile systems needs to be encapsulated in modules, in order to create a simple overall structure enabling robustness analysis. The proposed design is illustrated using a typical AUV mission.

In many multirobot applications, planning trajectories in a way to guarantee that the collective behavior of the robots satisfies a certain high-level specification is crucial. Motivated by this problem, we introduce counting temporal logics---formal languages that enable concise expression of multirobot task specifications over possibly infinite horizons. We first introduce a general logic called counting linear temporal logic plus (cLTL+), and propose an optimization-based method that generates individual trajectories such that satisfaction of a given cLTL+ formula is guaranteed when these trajectories are synchronously executed. We then introduce a fragment of cLTL+, called counting linear temporal logic (cLTL), and show that a solution to planning problem with cLTL constraints can be obtained more efficiently if all robots have identical dynamics. In the second part of the paper, we relax the synchrony assumption and discuss how to generate trajectories that can be asynchronously executed, while preserving the satisfaction of the desired cLTL+ specification. In particular, we show that when the asynchrony between robots is bounded, the method presented in this paper can be modified to generate robust trajectories. We demonstrate these ideas with an experiment and provide numerical results that showcase the scalability of the method.

In this paper, we consider the problem of formally verifying the safety of an autonomous robot equipped with a Neural Network (NN) controller that processes LiDAR images to produce control actions. Given a workspace that is characterized by a set of polytopic obstacles, our objective is to compute the set of safe initial conditions such that a robot trajectory starting from these initial conditions is guaranteed to avoid the obstacles. Our approach is to construct a finite state abstraction of the system and use standard reachability analysis over the finite state abstraction to compute the set of the safe initial states. The first technical problem in computing the finite state abstraction is to mathematically model the imaging function that maps the robot position to the LiDAR image. To that end, we introduce the notion of imaging-adapted sets as partitions of the workspace in which the imaging function is guaranteed to be affine. We develop a polynomial-time algorithm to partition the workspace into imaging-adapted sets along with computing the corresponding affine imaging functions. Given this workspace partitioning, a discrete-time linear dynamics of the robot, and a pre-trained NN controller with Rectified Linear Unit (ReLU) nonlinearity, the second technical challenge is to analyze the behavior of the neural network. To that end, we utilize a Satisfiability Modulo Convex (SMC) encoding to enumerate all the possible segments of different ReLUs. SMC solvers then use a Boolean satisfiability solver and a convex programming solver and decompose the problem into smaller subproblems. To accelerate this process, we develop a pre-processing algorithm that could rapidly prune the space feasible ReLU segments. Finally, we demonstrate the efficiency of the proposed algorithms using numerical simulations with increasing complexity of the neural network controller.

This paper presents iterative Sequential Action Control (iSAC), a receding horizon approach for control of nonlinear systems. The iSAC method has a closed-form open-loop solution, which is iteratively updated between time steps by introducing constant control values applied for short duration. Application of a contractive constraint on the cost is shown to lead to closed-loop asymptotic stability under mild assumptions. The effect of asymptotically decaying disturbances on system trajectories is also examined. To demonstrate the applicability of iSAC to a variety of systems and conditions, we employ five different systems, including a 13-dimensional quaternion-based quadrotor. Each system is tested in different scenarios, ranging from feasible and infeasible trajectory tracking, to setpoint stabilization, with or without the presence of external disturbances. Finally, limitations of this work are discussed.

We consider the off-policy estimation problem of estimating the expected reward of a target policy using samples collected by a different behavior policy. Importance sampling (IS) has been a key technique to derive (nearly) unbiased estimators, but is known to suffer from an excessively high variance in long-horizon problems. In the extreme case of in infinite-horizon problems, the variance of an IS-based estimator may even be unbounded. In this paper, we propose a new off-policy estimation method that applies IS directly on the stationary state-visitation distributions to avoid the exploding variance issue faced by existing estimators.Our key contribution is a novel approach to estimating the density ratio of two stationary distributions, with trajectories sampled from only the behavior distribution. We develop a mini-max loss function for the estimation problem, and derive a closed-form solution for the case of RKHS. We support our method with both theoretical and empirical analyses.

Collecting training data from the physical world is usually time-consuming and even dangerous for fragile robots, and thus, recent advances in robot learning advocate the use of simulators as the training platform. Unfortunately, the reality gap between synthetic and real visual data prohibits direct migration of the models trained in virtual worlds to the real world. This paper proposes a modular architecture for tackling the virtual-to-real problem. The proposed architecture separates the learning model into a perception module and a control policy module, and uses semantic image segmentation as the meta representation for relating these two modules. The perception module translates the perceived RGB image to semantic image segmentation. The control policy module is implemented as a deep reinforcement learning agent, which performs actions based on the translated image segmentation. Our architecture is evaluated in an obstacle avoidance task and a target following task. Experimental results show that our architecture significantly outperforms all of the baseline methods in both virtual and real environments, and demonstrates a faster learning curve than them. We also present a detailed analysis for a variety of variant configurations, and validate the transferability of our modular architecture.

Unmanned Aerial Vehicles (UAVs) have recently rapidly grown to facilitate a wide range of innovative applications that can fundamentally change the way cyber-physical systems (CPSs) are designed. CPSs are a modern generation of systems with synergic cooperation between computational and physical potentials that can interact with humans through several new mechanisms. The main advantages of using UAVs in CPS application is their exceptional features, including their mobility, dynamism, effortless deployment, adaptive altitude, agility, adjustability, and effective appraisal of real-world functions anytime and anywhere. Furthermore, from the technology perspective, UAVs are predicted to be a vital element of the development of advanced CPSs. Therefore, in this survey, we aim to pinpoint the most fundamental and important design challenges of multi-UAV systems for CPS applications. We highlight key and versatile aspects that span the coverage and tracking of targets and infrastructure objects, energy-efficient navigation, and image analysis using machine learning for fine-grained CPS applications. Key prototypes and testbeds are also investigated to show how these practical technologies can facilitate CPS applications. We present and propose state-of-the-art algorithms to address design challenges with both quantitative and qualitative methods and map these challenges with important CPS applications to draw insightful conclusions on the challenges of each application. Finally, we summarize potential new directions and ideas that could shape future research in these areas.

Due to their interdisciplinary nature, devices for controlled-environment agriculture have the possibility to turn into ideal tools not only to conduct research on plant phenology but also to create curricula in a wide range of disciplines. Controlled-environment devices are increasing their functionalities as well as improving their accessibility. Traditionally, building one of these devices from scratch implies knowledge in fields such as mechanical engineering, digital electronics, programming, and energy management. However, the requirements of an effective controlled environment device for personal use brings new constraints and challenges. This paper presents the OpenAg Personal Food Computer (PFC); a low cost desktop size platform, which not only targets plant phenology researchers but also hobbyists, makers, and teachers from elementary to high-school levels (K-12). The PFC is completely open-source and it is intended to become a tool that can be used for collective data sharing and plant growth analysis. Thanks to its modular design, the PFC can be used in a large spectrum of activities.

This paper explores the capability of deep neural networks to capture key characteristics of vehicle dynamics, and their ability to perform coupled longitudinal and lateral control of a vehicle. To this extent, two different artificial neural networks are trained to compute vehicle controls corresponding to a reference trajectory, using a dataset based on high-fidelity simulations of vehicle dynamics. In this study, control inputs are chosen as the steering angle of the front wheels, and the applied torque on each wheel. The performance of both models, namely a Multi-Layer Perceptron (MLP) and a Convolutional Neural Network (CNN), is evaluated based on their ability to drive the vehicle on a challenging test track, shifting between long straight lines and tight curves. A comparison to conventional decoupled controllers on the same track is also provided.

We introduce a stochastic model of diffeomorphisms, whose action on a variety of data types descends to stochastic evolution of shapes, images and landmarks. The stochasticity is introduced in the vector field which transports the data in the Large Deformation Diffeomorphic Metric Mapping (LDDMM) framework for shape analysis and image registration. The stochasticity thereby models errors or uncertainties of the flow in following the prescribed deformation velocity. The approach is illustrated in the example of finite dimensional landmark manifolds, whose stochastic evolution is studied both via the Fokker-Planck equation and by numerical simulations. We derive two approaches for inferring parameters of the stochastic model from landmark configurations observed at discrete time points. The first of the two approaches matches moments of the Fokker-Planck equation to sample moments of the data, while the second approach employs an Expectation-Maximisation based algorithm using a Monte Carlo bridge sampling scheme to optimise the data likelihood. We derive and numerically test the ability of the two approaches to infer the spatial correlation length of the underlying noise.

Policy gradient methods are an appealing approach in reinforcement learning because they directly optimize the cumulative reward and can straightforwardly be used with nonlinear function approximators such as neural networks. The two main challenges are the large number of samples typically required, and the difficulty of obtaining stable and steady improvement despite the nonstationarity of the incoming data. We address the first challenge by using value functions to substantially reduce the variance of policy gradient estimates at the cost of some bias, with an exponentially-weighted estimator of the advantage function that is analogous to TD(lambda). We address the second challenge by using trust region optimization procedure for both the policy and the value function, which are represented by neural networks. Our approach yields strong empirical results on highly challenging 3D locomotion tasks, learning running gaits for bipedal and quadrupedal simulated robots, and learning a policy for getting the biped to stand up from starting out lying on the ground. In contrast to a body of prior work that uses hand-crafted policy representations, our neural network policies map directly from raw kinematics to joint torques. Our algorithm is fully model-free, and the amount of simulated experience required for the learning tasks on 3D bipeds corresponds to 1-2 weeks of real time.

Given a sequential learning algorithm and a target model, sequential machine teaching aims to find the shortest training sequence to drive the learning algorithm to the target model. We present the first principled way to find such shortest training sequences. Our key insight is to formulate sequential machine teaching as a time-optimal control problem. This allows us to solve sequential teaching by leveraging key theoretical and computational tools developed over the past 60 years in the optimal control community. Specifically, we study the Pontryagin Maximum Principle, which yields a necessary condition for optimality of a training sequence. We present analytic, structural, and numerical implications of this approach on a case study with a least-squares loss function and gradient descent learner. We compute optimal training sequences for this problem, and although the sequences seem circuitous, we find that they can vastly outperform the best available heuristics for generating training sequences.

Many unmanned aerial vehicle surveillance and monitoring applications require observations at precise locations over long periods of time, ideally days or weeks at a time (e.g. ecosystem monitoring), which has been impractical due to limited endurance and the requirement of humans in the loop for operation. To overcome these limitations, we propose a fully autonomous small rotorcraft UAS that is capable of performing repeated sorties for long-term observation missions without any human intervention. We address two key technologies that are critical for such a system: full platform autonomy including emergency response to enable mission execution independently from human operators, and the ability of vision-based precision landing on a recharging station for automated energy replenishment. Experimental results of up to 11 hours of fully autonomous operation in indoor and outdoor environments illustrate the capability of our system.

Inverse problems appear in many applications, such as image deblurring and inpainting. The common approach to address them is to design a specific algorithm for each problem. The Plug-and-Play (P&P) framework, which has been recently introduced, allows solving general inverse problems by leveraging the impressive capabilities of existing denoising algorithms. While this fresh strategy has found many applications, a burdensome parameter tuning is often required in order to obtain high-quality results. In this work, we propose an alternative method for solving inverse problems using off-the-shelf denoisers, which requires less parameter tuning. First, we transform a typical cost function, composed of fidelity and prior terms, into a closely related, novel optimization problem. Then, we propose an efficient minimization scheme with a plug-and-play property, i.e., the prior term is handled solely by a denoising operation. Finally, we present an automatic tuning mechanism to set the method's parameters. We provide a theoretical analysis of the method, and empirically demonstrate its competitiveness with task-specific techniques and the P&P approach for image inpainting and deblurring.

Decades of research in control theory have shown that simple controllers, when provided with timely feedback, can control complex systems. Pushing is an example of a complex mechanical system that is difficult to model accurately due to unknown system parameters such as coefficients of friction and pressure distributions. In this paper, we explore the data-complexity required for controlling, rather than modeling, such a system. Results show that a model-based control approach, where the dynamical model is learned from data, is capable of performing complex pushing trajectories with a minimal amount of training data (10 data points). The dynamics of pushing interactions are modeled using a Gaussian process (GP) and are leveraged within a model predictive control approach that linearizes the GP and imposes actuator and task constraints for a planar manipulation task.

This paper presents a novel regrasp control policy that makes use of tactile sensing to plan local grasp adjustments. Our approach determines regrasp actions by virtually searching for local transformations of tactile measurements that improve the quality of the grasp. First, we construct a tactile-based grasp quality metric using a deep convolutional neural network trained on over 2800 grasps. The quality of each grasp, a continuous value between 0 and 1, is determined experimentally by measuring its resistance to external perturbations. Second, we simulate the tactile imprints associated with robot motions relative to the initial grasp by performing rigid-body transformations of the given tactile measurements. The newly generated tactile imprints are evaluated with the learned grasp quality network and the regrasp action is chosen to maximize the grasp quality. Results show that the grasp quality network can predict the outcome of grasps with an average accuracy of 85% on known objects and 75% on a cross validation set of 12 objects. The regrasp control policy improves the success rate of grasp actions by an average relative increase of 70% on a test set of 8 objects.

Noisy observations coupled with nonlinear dynamics pose one of the biggest challenges in robot motion planning. By decomposing nonlinear dynamics into a discrete set of local dynamics models, hybrid dynamics provide a natural way to model nonlinear dynamics, especially in systems with sudden discontinuities in dynamics due to factors such as contacts. We propose a hierarchical POMDP planner that develops cost-optimized motion plans for hybrid dynamics models. The hierarchical planner first develops a high-level motion plan to sequence the local dynamics models to be visited and then converts it into a detailed continuous state plan. This hierarchical planning approach results in a decomposition of the POMDP planning problem into smaller sub-parts that can be solved with significantly lower computational costs. The ability to sequence the visitation of local dynamics models also provides a powerful way to leverage the hybrid dynamics to reduce state uncertainty. We evaluate the proposed planner on a navigation task in the simulated domain and on an assembly task with a robotic manipulator, showing that our approach can solve tasks having high observation noise and nonlinear dynamics effectively with significantly lower computational costs compared to direct planning approaches.

In this paper, we present a whole-body control framework for Wheeled Inverted Pendulum (WIP) Humanoids. WIP Humanoids are redundant manipulators dynamically balancing themselves on wheels. Characterized by several degrees of freedom, they have the ability to perform several tasks simultaneously, such as balancing, maintaining a body pose, controlling the gaze, lifting a load or maintaining end-effector configuration in operation space. The problem of whole-body control is to enable simultaneous performance of these tasks with optimal participation of all degrees of freedom at specified priorities for each objective. The control also has to obey constraint of angle and torque limits on each joint. The proposed approach is hierarchical with a low level controller for body joints manipulation and a high-level controller that defines center of mass (CoM) targets for the low-level controller to control zero dynamics of the system driving the wheels. The low-level controller plans for shorter horizons while considering more complete dynamics of the system, while the high-level controller plans for longer horizon based on an approximate model of the robot for computational efficiency.

Complex motions for robots are frequently generated by switching among a collection of individual movement primitives. We use this approach to formulate robot motion plans as sequences of primitives to be executed one after the other. When dealing with dynamical movement primitives, besides accomplishing the high-level objective, planners must also reason about the effect of the plan's execution on the safety of the platform. This task becomes more daunting in the presence of disturbances, such as external forces. To alleviate this issue, we present a framework that builds on rigorous control-theoretic tools to generate safely-executable motion plans for externally excited robotic systems. Our framework is illustrated on a 3D limit-cycle gait bipedal robot that adapts its walking pattern to persistent external forcing.

Robots and autonomous agents often complete goal-based tasks with limited resources, relying on imperfect models and sensor measurements. In particular, reinforcement learning (RL) and feedback control can be used to help a robot achieve a goal. Taking advantage of this body of work, this paper formulates general computation as a feedback-control problem, which allows the agent to autonomously overcome some limitations of standard procedural language programming: resilience to errors and early program termination. Our formulation considers computation to be trajectory generation in the program's variable space. The computing then becomes a sequential decision making problem, solved with reinforcement learning (RL), and analyzed with Lyapunov stability theory to assess the agent's resilience and progression to the goal. We do this through a case study on a quintessential computer science problem, array sorting. Evaluations show that our RL sorting agent makes steady progress to an asymptotically stable goal, is resilient to faulty components, and performs less array manipulations than traditional Quicksort and Bubble sort.

This paper presents an approach to damp out the oscillatory motion of the pendulum-like hanging platform on which a robotic manipulator is mounted. To this end, moving masses were installed on top of the platform. In this paper, asymptotic stability of the platform (which implies oscillation damping) is achieved by designing reference acceleration of the moving masses properly. A main feature of this work is that we can achieve asymptotic stability of not only the platform, but also the moving masses, which may be challenging due to the under-actuation nature. The proposed scheme is validated by the simulation studies.

The objective is to study an on-line Hidden Markov model (HMM) estimation-based Q-learning algorithm for partially observable Markov decision process (POMDP) on finite state and action sets. When the full state observation is available, Q-learning finds the optimal action-value function given the current action (Q function). However, Q-learning can perform poorly when the full state observation is not available. In this paper, we formulate the POMDP estimation into a HMM estimation problem and propose a recursive algorithm to estimate both the POMDP parameter and Q function concurrently. Also, we show that the POMDP estimation converges to a set of stationary points for the maximum likelihood estimate, and the Q function estimation converges to a fixed point that satisfies the Bellman optimality equation weighted on the invariant distribution of the state belief determined by the HMM estimation process.

We present two distributed methods for the estimation of the kinematic parameters, the dynamic parameters, and the kinematic state of an unknown planar body manipulated by a decentralized multi-agent system. The proposed approaches rely on the rigid body kinematics and dynamics, on nonlinear observation theory, and on consensus algorithms. The only three requirements are that each agent can exert a 2D wrench on the load, it can measure the velocity of its contact point, and that the communication graph is connected. Both theoretical nonlinear observability analysis and convergence proofs are provided. The first method assumes constant parameters while the second one can deal with time-varying parameters and can be applied in parallel to any task-oriented control law. For the cases in which a control law is not provided, we propose a distributed and safe control strategy satisfying the observability condition. The effectiveness and robustness of the estimation strategy is showcased by means of realistic MonteCarlo simulations.

Drone delivery has been a hot topic in the industry in the past few years. However, existing approaches either focus on rural areas or rely on centralized drop-off locations from where the last mile delivery is performed. In this paper we tackle the problem of autonomous last mile delivery in urban environments using an off-the-shelf drone. We build a prototype system that is able to fly to the approximate delivery location using GPS and then find the exact drop-off location using visual navigation. The drop-off location could, e.g., be on a balcony or porch, and simply needs to be indicated by a visual marker on the wall or window. We test our system components in simulated environments, including the visual navigation and collision avoidance. Finally, we deploy our drone in a real-world environment and show how it can find the drop-off point on a balcony. To stimulate future research in this topic we open source our code.

We present Flipper, a natural language interface for describing high-level task specifications for robots that are compiled into robot actions. Flipper starts with a formal core language for task planning that allows expressing rich temporal specifications and uses a semantic parser to provide a natural language interface. Flipper provides immediate visual feedback by executing an automatically constructed plan of the task in a graphical user interface. This allows the user to resolve potentially ambiguous interpretations. Flipper extends itself via naturalization: its users can add definitions for utterances, from which Flipper induces new rules and adds them to the core language, gradually growing a more and more natural task specification language. Flipper improves the naturalization by generalizing the definition provided by users. Unlike other task-specification systems, Flipper enables natural language interactions while maintaining the expressive power and formal precision of a programming language. We show through an initial user study that natural language interactions and generalization can considerably ease the description of tasks. Moreover, over time, users employ more and more concepts outside of the initial core language. Such extensions are available to the Flipper community, and users can use concepts that others have defined.