IROS 2022 论文列表

In this study, a mechanism that realizes a novel structural form of the harmonic reducer is introduced. Conventional robots often use various mechanical reducers owing to low torque and high-speed characteristics of electric motors. Among them, harmonic reducers are frequently used because of their compact size and backlash-free precision. The plate harmonic reducer which uses the same topological geometry and reducing mechanism as the conventional harmonic reducer is a novel type of strain gear that changes its shape to a plate form for axial deformation. It has unique differences in terms of axial thickness, torsional stiffness, and efficiency due to its morphological characteristics. This study introduces and analyzes the reducing principle of the plate harmonic reducer and describes the methodological solutions for realization. Finally, the theoretical performance improvement and operating feasibility of the plate harmonic reducer are analyzed using finite element method and a 3D-printed prototype model.

Contemporary planning models and methods often rely on constant availability of free state information at each step of execution. However, autonomous systems are increasingly deployed in the open world where state information may be costly or simply unavailable in certain situations. Failing to account for sensor limitations may lead to costly behavior or even catastrophic failure. While the partially observable Markov decision process (POMDP) can be used to model this problem, solving POMDPs is often intractable. We introduce a planning model called a semi-observable Markov decision process (SOMDP) specifically designed for MDPs where state observability may be intermittent. We propose an approach for solving SOMDPs that uses memory states to proactively plan for the potential loss of sensor information while exploiting the unique structure of SOMDPs. Our theoretical analysis and empirical evaluation demonstrate the advantages of SOMDPs relative to existing planning models.

We study planning problems faced by robots operating in uncertain environments with incomplete knowledge of state, and actions that are noisy and/or imprecise. This paper identifies a new problem sub-class that models settings in which information is revealed only intermittently through some exogenous process that provides state information periodically. Several practical domains fit this model, including the specific scenario that motivates our research: autonomous navigation of a planetary exploration rover augmented by remote imaging. With an eye to efficient specialized solution methods, we examine the structure of instances of this sub-class. They lead to Markov Decision Processes with exponentially large action-spaces but for which, as those actions comprise sequences of more atomic elements, one may establish performance bounds by comparing policies under different information assumptions. This provides a way in which to construct performance bounds systematically. Such bounds are useful because, in conjunction with the insights they confer, they can be employed in bounding-based methods to obtain high-quality solutions efficiently; the empirical results we present demonstrate their effectiveness for the considered problems. The foregoing has also alluded to the distinctive role that time plays for these problems -more specifically: time until information is revealed- and we uncover and discuss several interesting subtleties in this regard.

General-purpose robots require diverse repertoires of behaviors to complete challenging tasks in real-world unstructured environments. To address this issue, goal-conditioned reinforcement learning aims to acquire policies that can reach configurable goals for a wide range of tasks on command. However, such goal-conditioned policies are notoriously difficult and time-consuming to train from scratch. In this paper, we propose Planning to Practice (PTP), a method that makes it practical to train goal-conditioned policies for long-horizon tasks that require multiple distinct types of interactions to solve. Our approach is based on two key ideas. First, we decompose the goal-reaching problem hierarchically, with a high-level planner that sets intermediate subgoals using conditional subgoal generators in the latent space for a low-level model-free policy. Second, we propose a hybrid approach which first pre-trains both the conditional subgoal generator and the policy on previously collected data through offline reinforcement learning, and then fine-tunes the policy via online exploration. This fine-tuning process is itself facilitated by the planned subgoals, which breaks down the original target task into short-horizon goal-reaching tasks that are significantly easier to learn. We conduct experiments in both the simulation and real world, in which the policy is pre-trained on demonstrations of short primitive behaviors and fine-tuned for temporally extended tasks that are unseen in the offline data. Our experimental results show that PTP can generate feasible sequences of subgoals that enable the policy to efficiently solve the target tasks. 11Supplementary video: sites.google.com/view/planning-to-practice

Planning autonomous driving behaviors in dense traffic is challenging. Human drivers are able to influence their road environment to achieve (otherwise unachievable) goals, by communicating their intents to other drivers. An autonomous system that is required to drive in the presence of human traffic must thus possess this fundamental negotiation capability. This work presents a novel benchmark that includes a stochastic driver negotiation model and a framework for training policies to drive and negotiate based on reinforcement learning. It is shown that driving policies trained in this framework lead to greater safety, higher mission accomplishment rates and more driving comfort, and can generalize across scenarios.

In this paper, we consider a multi-sensor arrangement of proximity sensors that forms a proximity array. A general modeling methodology is considered within the framework of Sensor-based Control. It incorporates multiple sensor signals from the proximity array by giving primary emphasis on the interaction screw. To prove its effectiveness, modeling approach is applied to the task of plane-to-plane positioning. We discuss the development of two sensor-based task functions for the specific task considered. The validity of the methodology is provided using relevant experimental results.

Soft robotics has the potential to revolutionize robotic locomotion, in particular, soft robotic swimmers offer a minimally invasive and adaptive solution to explore and preserve our oceans. Unfortunately, current soft robotic swimmers are vastly inferior to evolved biological swimmers, especially in terms of controllability, efficiency, maneuverability, and longevity. Additionally, the tedious iterative fabrication and empirical testing required to design soft robots has hindered their optimization. In this work, we tackle this challenge by providing an efficient and straightforward pipeline for designing and fabricating soft robotic swimmers equipped with electrostatic actuation. We streamline the process to allow for rapid additive manufacturing, and show how a differentiable simulation can be used to match a simplified model to the real deformation of a robotic swimmer. We perform several experiments with the fabricated swimmer by varying the voltage and actuation frequency of the swimmer's antagonistic muscles. We show how the voltage and frequency vary the locomotion speed of the swimmer while moving in liquid oil and observe a clear optimum in forward swimming speed. The differentiable simulation model we propose has various downstream applications, such as control and shape optimization of the swimmer; optimization results can be directly mapped back to the real robot through our sim-to-real matching.

Non-holonomic vehicle motion has been studied extensively using physics-based models. Common approaches when using these models interpret the wheel/ground interactions using a linear tire model and thus may not fully capture the nonlinear and complex dynamics under various environments. On the other hand, neural network models have been widely employed in this domain, demonstrating powerful function approximation capabilities. However, these black-box learning strategies completely abandon the existing knowledge of well-known physics. In this paper, we seamlessly combine deep learning with a fully differentiable physics model to endow the neural network with available prior knowledge. The proposed model shows better generalization performance than the vanilla neural network model by a large margin. We also show that the latent features of our model can accurately represent lateral tire forces without the need for any additional training. Lastly, We develop a risk-aware model predictive controller using proprioceptive information derived from the latent features. We validate our idea in two autonomous driving tasks under unknown friction, outperforming the baseline control framework.

Materials with the ability to self-classify their own shape have the potential to advance a wide range of engineering applications and industries. Biological systems possess the ability not only to self-reconfigure but also to self-classify themselves to determine a general shape and function. Previous work into modular robotics systems has only enabled self-recognition and self-reconfiguration into a specific target shape, missing the inherent robustness present in nature to self-classify. In this paper we therefore take advantage of recent advances in deep learning and neural cellular automata, and present a simple modular 2D robotic system that can infer its own class of shape through the local communication of its components. Furthermore, we show that our system can be successfully transferred to hardware which thus opens op-portunities for future self-classifying machines. Code available at https://github.com/kattwalker/projectcube. Video available at https://youtu.be/0TCOkE4keyc.

Visual SLAM inside the human body will open the way to computer-assisted navigation in endoscopy. However, due to space limitations, medical endoscopes only provide monocular images, leading to systems lacking true scale. In this paper, we exploit the controlled lighting in colonoscopy to achieve the first in-vivo 3D reconstruction of the human colon using photometric stereo on a calibrated monocular endoscope. Our method works in a real medical environment, providing both a suitable in-place calibration procedure and a depth estimation technique adapted to the colon's tubular geometry. We validate our method on simulated colonoscopies, obtaining a mean error of 7% on depth estimation, which is below 3 mm on average. Our qualitative results on the EndoMapper dataset show that the method is able to correctly estimate the colon shape in real human colonoscopies, paving the ground for truescale monocular SLAM in endoscopy.

Despite non-co-location, haptic stimulation at the wrist can potentially provide feedback regarding interactions at the fingertips without encumbering the user's hand. Here we investigate how two types of skin deformation at the wrist (normal and shear) relate to the perception of the mechanical properties of virtual objects. We hypothesized that a congruent mapping (i.e. when the most relevant interaction forces during a virtual interaction spatially match the haptic feedback at the wrist) would result in better perception than other mappings. We performed an experiment where haptic devices at the wrist rendered either normal or shear feedback during manipulation of virtual objects with varying stiffness, mass, or friction properties. Perception of mechanical properties was more accurate with congruent skin stimulation than noncongruent. In addition, discrimination performance and subjective reports were positively influenced by congruence. This study demonstrates that users can perceive mechanical properties via haptic feedback provided at the wrist with a consistent mapping between haptic feedback and interaction forces at the fingertips, regardless of congruence.

We present Perceive-Represent-Generate (PRG), a novel three-stage framework that maps perceptual information of different modalities (e.g., visual or sound), corresponding to a series of instructions, to a sequence of movements to be executed by a robot. In the first stage, we perceive and preprocess the given inputs, isolating individual commands from the complete instruction provided by a human user. In the second stage we encode the individual commands into a multimodal latent space, employing a deep generative model. Finally, in the third stage we convert the latent samples into individual trajectories and combine them into a single dynamic movement primitive, allowing its execution by a robotic manipulator. We evaluate our pipeline in the context of a novel robotic handwriting task, where the robot receives as input a word through different perceptual modalities (e.g., image, sound), and generates the corresponding motion trajectory to write it, creating coherent and high-quality handwritten words.

Autonomous navigation in highly populated areas remains a challenging task for robots because of the difficulty in guaranteeing safe interactions with pedestrians in unstructured situations. In this work, we present a crowd navigation control framework that delivers continuous obstacle avoidance and post-contact control evaluated on an autonomous personal mobility vehicle. We propose evaluation metrics for accounting efficiency, controller response and crowd interactions in natural crowds. We report the results of over 110 trials in different crowd types: sparse, flows, and mixed traffic, with low- (< 0.15 ppsm), mid- (< 0.65 ppsm), and high- (< 1 ppsm) pedestrian densities. We present comparative results between two low-level obstacle avoidance methods and a baseline of shared control. Results show a 10% drop in relative time to goal on the highest density tests, and no other efficiency metric decrease. Moreover, autonomous navigation showed to be comparable to shared-control navigation with a lower relative jerk and significantly higher fluency in commands indicating high compatibility with the crowd. We conclude that the reactive controller fulfils a necessary task of fast and continuous adaptation to crowd navigation, and it should be coupled with high-level planners for environmental and situational awareness.

Anticipating the future behavior of pedestrians is a crucial part of deploying Automated Driving Systems (ADS) in urban traffic scenarios. Most recent works utilize a convolutional neural network (CNN) to extract visual information, which is then input to a recurrent neural network (RNN) along with pedestrian-specific features like location and speed to obtain temporal features. However, the majority of these approaches lack the ability to parse the relationships of the related objects in the specific traffic scene, which leads to omitting the interactions between the pedestrians and the interactions between the pedestrians and the traffic. For this purpose, we propose a graph-structured model which can dig out pedestrians' dynamic constraints by constructing a traffic-aware scene graph within each frame. In addition, to capture pedestrian movement more effectively, we also introduce a temporal feature representation model, which first uses inter-frame and intra-frame GRU (II-GRU) to mine inter-frame information and intra-frame information together, and then employs a novel attention mechanism to adaptively generate attention weights. Extensive experiments on the JAAD and PIE datasets prove that our proposed model is effective in reaching and enhancing the state-of-the-art performance.

In the field of 3D perception using 3D LiDAR sensors, ground segmentation is an essential task for various purposes, such as traversable area detection and object recognition. Under these circumstances, several ground segmentation methods have been proposed. However, some limitations are still encountered. First, some ground segmentation methods require fine-tuning of parameters depending on the surroundings, which is excessively laborious and time-consuming. Moreover, even if the parameters are well adjusted, a partial under-segmentation problem can still emerge, which implies ground segmentation failures in some regions. Finally, ground segmentation methods typically fail to estimate an appropriate ground plane when the ground is above another structure, such as a retaining wall. To address these problems, we propose a robust ground segmentation method called Patchwork++, an extension of Patchwork. Patchwork++ exploits adaptive ground likelihood estimation (A-GLE) to calculate appropriate parameters adaptively based on the previous ground segmentation results. Moreover, temporal ground revert (TGR) alleviates a partial under-segmentation problem by using the temporary ground property. Also, region-wise vertical plane fitting (R-VPF) is introduced to segment the ground plane properly even if the ground is elevated with different layers. Finally, we present reflected noise removal (RNR) to eliminate virtual noise points efficiently based on the 3D LiDAR reflection model. We demonstrate the qualitative and quantitative evaluations using a SemanticKITTI dataset. Our code is available at https://github.com/url-kaist/patchwork-plusplus

Tactile robots shall be deployed for dynamic task execution in production lines with small batch sizes. Therefore, these robots should have the ability to respond to changing conditions and be easy to (re-)program. Operating under uncertain environments requires unifying subsystems such as robot motion and force policy into one framework, referred to as tactile skills. In this paper, we propose the enhancement of these skills for passivity-based skill motion learning in stiffness-adaptive unified force-impedance control. To achieve the increased level of adaptability, we represent all tactile skills by three basic primitives: contact initiation, manipulation, and contact termination. To ensure passivity and stability, we develop an energy-based approach for unified force-impedance control that allows humans to teach the robot motion through physical interaction during the execution of a tactile task. We incorporate our proposed framework into a tactile robot to experimentally validate the motion adaptation by interaction performance and stability of the control. While the polishing task is presented as our use case through the paper, the experiments can also be carried out with various tactile skills. Finally, the results show the novel controller's stability and passivity to contact-loss and stiffness adaptation, leading to successful programming by interaction.

To deploy robots outside of laboratory environments, they must be able to locomote on natural, unstructured terrain. While perception and control strategies for terrain navigation and obstacle avoidance have been developed for human-scale robots, microrobots are often too small to carry the sensors, computing power, and energy required to implement such techniques. Instead, this work presents passive foot designs for improving open-loop, quasi-static locomotion of a 1.6 g, 45 mm quadruped robot over rough terrains. The feet were evaluated by tracking the distance travelled on an uneven terrain of progressively increasing feature heights. Our insect tarsi inspired rigid foot designs improved performance somewhat, and by adding passive compliance via a viscoelastic hinge on the heel and toe, we increased the distance travelled by 168% over the original design. By exploring the design space of foot geometries and compliance, this work lays the foundation for understanding how passive foot design facilitates locomotion over uneven terrains.

An environment where a robot swarm attacks a territory protected by another one leads to an attack-defense confrontation problem. Commonly-used deep reinforcement learning-based methods rely on pre-training and become intractable due to the curse of dimensionality. To develop effective attack strategies, inspired by a particle swarm optimizer (PSO), this work proposes a PSO-based strategy for a robot swarm for the first time. During the moving of a robot swarm, each robot obtains situation information through perceiving its nearby peers and enemies and uses such information to construct its fitness function. Then, each robot uses PSO to optimize its fitness function and searches for its optimal attack position, which guides it to move in the next time slot. The experimental analyses show that the PSO-based attack strategy has more potential in solving large-scale confrontational problems than the deep reinforcement learning-based algorithms.

We propose a novel Parallel Monte Carlo tree search with Batched Simulations (PMBS) algorithm for accelerating long-horizon, episodic robotic planning tasks. Monte Carlo tree search (MCTS) is an effective heuristic search algorithm for solving episodic decision-making problems whose underlying search spaces are expansive. Leveraging a GPU-based large-scale simulator, PMBS introduces massive parallelism into MCTS for solving planning tasks through the batched execution of a large number of concurrent simulations, which allows for more efficient and accurate evaluations of the expected cost-to-go over large action spaces. When applied to the challenging manipulation tasks of object retrieval from clutter, PMBS achieves a speedup of over 30× with an improved solution quality, in comparison to a serial MCTS implementation. We show that PMBS can be directly applied to real robot hardware with negligible sim-to-real differences. Supplementary material, including video, can be found at https://github.com/arc-l/pmbs.