This paper addresses a safe planning and control problem for mobile robots operating in communication- and sensor-limited dynamic environments. In this case the robots cannot sense the objects around them and must instead rely on intermittent, external information about the environment, as e.g., in underwater applications. The challenge in this case is that the robots must plan using only this stale data, while accounting for any noise in the data or uncertainty in the environment. To address this challenge we propose a compositional technique which leverages neural networks to quickly plan and control a robot through crowded and dynamic environments using only intermittent information. Specifically, our tool uses reachability analysis and potential fields to train a neural network that is capable of generating safe control actions. We demonstrate our technique both in simulation with an underwater vehicle crossing a crowded shipping channel and with real experiments with ground vehicles in communication- and sensor-limited environments.
相關內容
Discovering new intents is of great significance to establishing Bootstrapped Task-Oriented Dialogue System. Most existing methods either lack the ability to transfer prior knowledge in the known intent data or fall into the dilemma of forgetting prior knowledge in the follow-up. More importantly, these methods do not deeply explore the intrinsic structure of unlabeled data, so they can not seek out the characteristics that make an intent in general. In this paper, starting from the intuition that discovering intents could be beneficial to the identification of the known intents, we propose a probabilistic framework for discovering intents where intent assignments are treated as latent variables. We adopt Expectation Maximization framework for optimization. Specifically, In E-step, we conduct discovering intents and explore the intrinsic structure of unlabeled data by the posterior of intent assignments. In M-step, we alleviate the forgetting of prior knowledge transferred from known intents by optimizing the discrimination of labeled data. Extensive experiments conducted in three challenging real-world datasets demonstrate our method can achieve substantial improvements.
Motion planning and control are crucial components of robotics applications. Here, spatio-temporal hard constraints like system dynamics and safety boundaries (e.g., obstacles in automated driving) restrict the robot's motions. Direct methods from optimal control solve a constrained optimization problem. However, in many applications finding a proper cost function is inherently difficult because of the weighting of partially conflicting objectives. On the other hand, Imitation Learning (IL) methods such as Behavior Cloning (BC) provide a intuitive framework for learning decision-making from offline demonstrations and constitute a promising avenue for planning and control in complex robot applications. Prior work primarily relied on soft-constraint approaches, which use additional auxiliary loss terms describing the constraints. However, catastrophic safety-critical failures might occur in out-of-distribution (OOD) scenarios. This work integrates the flexibility of IL with hard constraint handling in optimal control. Our approach constitutes a general framework for constraint robotic motion planning and control using offline IL. Hard constraints are integrated into the learning problem in a differentiable manner, via explicit completion and gradient-based correction. Simulated experiments of mobile robot navigation and automated driving provide evidence for the performance of the proposed method.
Human awareness in robot motion planning is crucial for seamless interaction with humans. Many existing techniques slow down, stop, or change the robot's trajectory locally to avoid collisions with humans. Although using the information on the human's state in the path planning phase could reduce future interference with the human's movements and make safety stops less frequent, such an approach is less widespread. This paper proposes a novel approach to embedding a human model in the robot's path planner. The method explicitly addresses the problem of minimizing the path execution time, including slowdowns and stops owed to the proximity of humans. For this purpose, it converts safety speed limits into configuration-space cost functions that drive the path's optimization. The costmap can be updated based on the observed or predicted state of the human. The method can handle deterministic and probabilistic representations of the human state and is independent of the prediction algorithm. Numerical and experimental results on an industrial collaborative cell demonstrate that the proposed approach consistently reduces the robot's execution time and avoids unnecessary safety speed reductions.
When robots enter everyday human environments, they need to understand their tasks and how they should perform those tasks. To encode these, reward functions, which specify the objective of a robot, are employed. However, designing reward functions can be extremely challenging for complex tasks and environments. A promising approach is to learn reward functions from humans. Recently, several robot learning works embrace this approach and leverage human demonstrations to learn the reward functions. Known as inverse reinforcement learning, this approach relies on a fundamental assumption: humans can provide near-optimal demonstrations to the robot. Unfortunately, this is rarely the case: human demonstrations to the robot are often suboptimal due to various reasons, e.g., difficulty of teleoperation, robot having high degrees of freedom, or humans' cognitive limitations. This thesis is an attempt towards learning reward functions from human users by using other, more reliable data modalities. Specifically, we study how reward functions can be learned using comparative feedback, in which the human user compares multiple robot trajectories instead of (or in addition to) providing demonstrations. To this end, we first propose various forms of comparative feedback, e.g., pairwise comparisons, best-of-many choices, rankings, scaled comparisons; and describe how a robot can use these various forms of human feedback to infer a reward function, which may be parametric or non-parametric. Next, we propose active learning techniques to enable the robot to ask for comparison feedback that optimizes for the expected information that will be gained from that user feedback. Finally, we demonstrate the applicability of our methods in a wide variety of domains, ranging from autonomous driving simulations to home robotics, from standard reinforcement learning benchmarks to lower-body exoskeletons.
In this paper, we present a novel flow model and compensation strategy for high-viscosity fluid deposition that yields high quality parts in the face of large transient delays and nonlinearity. Robotic high-viscosity fluid deposition is an essential process for a broad range of manufacturing applications including additive manufacturing, adhesive and sealant dispensing, and soft robotics. However, high-viscosity fluid deposition without compensation can lead to poor part quality and defects due to large transient delays and complex fluid dynamics. Our computationally efficient model is well-suited to real-time control and can be quickly calibrated and our compensation strategy leverages an iterative Linear-Quadratic Regulator to compute compensated deposition paths that can be deployed on most dispensing systems, without additional hardware. We demonstrate the improvements provided by our method when 3D printing using a robotic manipulator.
The Internet is a critical resource in the day-to-day life of billions of users. To support the growing number of users and their increasing demands, operators have to continuously scale their network footprint -- e.g., by joining Internet Exchange Points -- and adopt relevant technologies -- such as IPv6. IPv6, however, has a vastly larger address space compared to its predecessor, which allows for new kinds of attacks on the Internet routing infrastructure. In this paper, we present Kirin: a BGP attack that sources millions of IPv6 routes and distributes them via thousands of sessions across various IXPs to overflow the memory of border routers within thousands of remote ASes. Kirin's highly distributed nature allows it to bypass traditional route-flooding defense mechanisms, such as per-session prefix limits or route flap damping. We analyze the theoretical feasibility of the attack by formulating it as a Integer Linear Programming problem, test for practical hurdles by deploying the infrastructure required to perform a small-scale Kirin attack using 4 IXPs, and validate our assumptions via BGP data analysis, real-world measurements, and router testbed experiments. Despite its low deployment cost, we find Kirin capable of injecting lethal amounts of IPv6 routes in the routers of thousands of ASes.
Urban road traffic continuously evolves under uncertainty. Existing traffic control systems only possess a local perspective over the multiple scales of traffic evolution, namely the intersection level, the corridor level, and the region level respectively. Capturing uncertainty under complex traffic spatio-temporal interactions is a very difficult problem and we often experience how fragile such systems are in reality. But luckily, despite its complex mechanics, traffic is described by various periodic phenomena. Workday flow distributions in the morning and evening commuting times can be exploited to make traffic adaptive and robust to disruptions. Additionally, controlling traffic is also based on a periodic process, choosing the phase of green time to allocate to opposite directions right of the pass and complementary red time phase for adjacent directions. In our work, we consider a novel system for road traffic control based on a network of interacting oscillators. Such a model has the advantage to capture temporal and spatial interactions of traffic light phasing as well as the network-level evolution of the traffic macroscopic features (i.e. flow, density). In this study, we propose a new realization of the antifragile control framework to control a network of interacting oscillator-based traffic light models to achieve region-level flow optimization. We demonstrate that antifragile control can capture the volatility of the urban road environment and the uncertainty about the distribution of the disruptions that can occur. We complement our control-theoretic design and analysis with experiments on a real-world setup comparatively discussing the benefits of an antifragile design for traffic control.
The question of answering queries over ML predictions has been gaining attention in the database community. This question is challenging because the cost of finding high quality answers corresponds to invoking an oracle such as a human expert or an expensive deep neural network model on every single item in the DB and then applying the query. We develop a novel unified framework for approximate query answering by leveraging a proxy to minimize the oracle usage of finding high quality answers for both Precision-Target (PT) and Recall-Target (RT) queries. Our framework uses a judicious combination of invoking the expensive oracle on data samples and applying the cheap proxy on the objects in the DB. It relies on two assumptions. Under the Proxy Quality assumption, proxy quality can be quantified in a probabilistic manner w.r.t. the oracle. This allows us to develop two algorithms: PQA that efficiently finds high quality answers with high probability and no oracle calls, and PQE, a heuristic extension that achieves empirically good performance with a small number of oracle calls. Alternatively, under the Core Set Closure assumption, we develop two algorithms: CSC that efficiently returns high quality answers with high probability and minimal oracle usage, and CSE, which extends it to more general settings. Our extensive experiments on five real-world datasets on both query types, PT and RT, demonstrate that our algorithms outperform the state-of-the-art and achieve high result quality with provable statistical guarantees.
We propose a framework for planning in unknown dynamic environments with probabilistic safety guarantees using conformal prediction. Particularly, we design a model predictive controller (MPC) that uses i) trajectory predictions of the dynamic environment, and ii) prediction regions quantifying the uncertainty of the predictions. To obtain prediction regions, we use conformal prediction, a statistical tool for uncertainty quantification, that requires availability of offline trajectory data - a reasonable assumption in many applications such as autonomous driving. The prediction regions are valid, i.e., they hold with a user-defined probability, so that the MPC is provably safe. We illustrate the results in the self-driving car simulator CARLA at a pedestrian-filled intersection. The strength of our approach is compatibility with state of the art trajectory predictors, e.g., RNNs and LSTMs, while making no assumptions on the underlying trajectory-generating distribution. To the best of our knowledge, these are the first results that provide valid safety guarantees in such a setting.
This paper presents a new multi-layered algorithm for motion planning under motion and sensing uncertainties for Linear Temporal Logic specifications. We propose a technique to guide a sampling-based search tree in the combined task and belief space using trajectories from a simplified model of the system, to make the problem computationally tractable. Our method eliminates the need to construct fine and accurate finite abstractions. We prove correctness and probabilistic completeness of our algorithm, and illustrate the benefits of our approach on several case studies. Our results show that guidance with a simplified belief space model allows for significant speed-up in planning for complex specifications.
小貼士
登錄享
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191