IROS 2022 论文列表

Autonomous navigation of ground robots has been widely used in indoor structured 2D environments, but there are still many challenges in outdoor 3D unstructured environments, especially in rough, uneven terrains. This paper proposed a plane-fitting based uneven terrain navigation framework (PUTN) to solve this problem. The implementation of PUTN is divided into three steps. First, based on Rapidly-exploring Random Trees (RRT), an improved sample-based algorithm called Plane Fitting RRT*(PF- RRT*) is proposed to obtain a sparse trajectory. Each sampling point corresponds to a custom traversability index and a fitted plane on the point cloud. These planes are connected in series to form a traversable “strip”. Second, Gaussian Process Regression is used to generate traversability of the dense trajectory interpolated from the sparse trajectory, and the sampling tree is used as the training set. Finally, local planning is performed using nonlinear model predictive control (NMPC). By adding the traversability index and uncertainty to the cost function, and adding obstacles generated by the real-time point cloud to the constraint function, a safe motion planning algorithm with smooth speed and strong robustness is available. Experiments in real scenarios are conducted to verify the effectiveness of the method. The source code is released for the reference of the community11Source code: https://github.com/jianzhuozhuTHU/putn..

Segmenting moving objects in the 3D LiDAR point cloud can provide important guidance to localization, mapping and decision-making for self-driving vehicles. As for the conventional approaches to point cloud segmentation, they rely on semantic-level information, which makes it inevitable for long-tail problems to arise as there are always unseen types of objects on the road. To achieve moving segmentation while avoiding the reliance on the object category, the point motion is identified in this paper by fully exploring and aggregating the point-level geometric consistency in sequential point clouds. More specifically, an end-to-end point-wise uncertainty weighted aggregation approach known as PUA-MOS is proposed to segment the moving points in 3D LiDAR Data. Our method is applicable to estimate point-wise moving mask, scene flow and rigid-body transformation simultaneously in a coarse- to-fine network, where the relations between each prediction are implicitly learned. To explicitly model the inner and inter relations across these predictions among all points, the point- wise estimation and the average value of the same motion points are aggregated according to a predicted uncertainty. Then, the aggregated estimation is fed again into the next-level fusion, where the points will be re-segmented using the aggregated mask from the last level. Through iterative joint aggregation, our PUA-MOS outperforms the previous methods significantly on both KITTI [4] and Waymo [26] datasets. The code will be provided to generate the moving segmentation labels on both datasets for reproduction.

Quantifying the safety of the human body ori-entation is an important issue in human-robot interaction. Knowing the changing physical constraints on human motion can improve inspection of safe human motions and bring essential information about stability and normality of human body orientations with real-time risk assessment. Also, this information can be used in cooperative robots and monitoring systems to evaluate and interact in the environment more freely. Furthermore, the workspace area can be more deterministic with the known physical characteristics of safety. Based on this motivation, we propose a novel predictive safety model (PSM) that relies on the information of an inertial measurement unit on the human chest. The PSM encompasses a 3-Dofs spring-damper pendulum model that predicts human motion based on a safe motion dataset. The estimated safe orientation of humans is obtained by integrating a safety dataset and an elastic spring-damper model in a way that the proposed approach can realize complex motions at different safety levels. We did experiments in a real-world scenario to verify our novel proposed model. This novel approach can be used in different guidance/assistive robots and health monitoring systems to support and evaluate the human condition, particularly elders.

Deep reinforcement learning (deep RL) has emerged as an effective tool for developing controllers for legged robots. However, vanilla deep RL often requires a tremendous amount of training samples and is not feasible for achieving robust behaviors. Instead, researchers have investigated a novel policy architecture by incorporating human experts' knowledge, such as Policies Modulating Trajectory Generators (PMTG). This architecture builds a recurrent control loop by combining a parametric trajectory generator (TG) and a feedback policy network to achieve more robust behaviors. In this work, we propose Policies Modulating Finite State Machine (PM-FSM) by replacing TGs with contact-aware finite state machines (FSM), which offers more flexible control of each leg. This invention offers an explicit notion of contact events to the policy to negotiate unexpected perturbations. We demonstrated that the proposed architecture could achieve more robust behaviors in various scenarios, such as challenging terrains or external perturbations, on both simulated and real robots.

Evolution Strategy (ES) algorithms have shown promising results in training complex robotic control policies due to their massive parallelism capability, simple implementation, effective parameter-space exploration, and fast training time. However, a key limitation of ES is its scalability to large capacity models, including modern neural network architectures. In this work, we develop Predictive Information Augmented Random Search (PI-ARS) to mitigate this limitation by leveraging recent advancements in representation learning to reduce the parameter search space for ES. Namely, PI-ARS combines a gradient-based representation learning technique, Predictive Information (PI), with a gradient-free ES algorithm, Augmented Random Search (ARS), to train policies that can process complex robot sensory inputs and handle highly nonlinear robot dynamics. We evaluate PI-ARS on a set of challenging visual-locomotion tasks where a quadruped robot needs to walk on uneven stepping stones, quincuncial piles, and moving platforms, as well as to complete an indoor navigation task. Across all tasks, PI-ARS demonstrates significantly better learning efficiency and performance compared to the ARS baseline. We further validate our algorithm by demonstrating that the learned policies can successfully transfer to a real quadruped robot, for example, achieving a 100% success rate on the real-world stepping stone environment, dramatically improving prior results achieving 40% success.

Simultaneous localization and mapping (SLAM) based on laser sensors has been widely adopted by mobile robots and autonomous vehicles. These SLAM systems are required to support accurate localization with limited computational resources. In particular, point cloud registration, i.e., the process of matching and aligning multiple LiDAR scans collected at multiple locations in a global coordinate framework, has been deemed as the bottleneck step in SLAM. In this paper, we propose a feature filtering algorithm, PFilter, that can filter out invalid features and can thus greatly alleviate this bottleneck. Meanwhile, the overall registration accuracy is also improved due to the carefully curated feature points. We integrate PFilter into the well-established scan-to-map LiDAR odometry framework, F-LOAM, and evaluate its performance on the KITTI dataset. The experimental results show that PFilter can remove about 48.4% of the points in the local feature map and reduce feature points in scan by 19.3% on average, which save 20.9% processing time per frame. In the mean time, we improve the accuracy by 9.4%.

Complexity, cost, and power requirements for the actuation of individual robots can play a large factor in limiting the size of robotic swarms. Here we present PCBot, a minimalist robot that can precisely move on an orbital shake table using a bi-stable solenoid actuator built directly into its PCB. This allows the actuator to be built as part of the automated PCB manufacturing process, greatly reducing the impact it has on manual assembly. Thanks to this novel actuator design, PCBot has merely five major components and can be assembled in under 20 seconds, potentially enabling them to be easily mass-manufactured. Here we present the electro-magnetic and mechanical design of PCBot. Additionally, a prototype robot is used to demonstrate its ability to move in a straight line as well as follow given paths.

In the narrow lane scene of autonomous driving, it is critical for the ego car to recognize the intentions of social vehicles and cooperate with them. However, cooperating with social vehicles is challenging due to insufficient information. This paper proposes an Explorative Game that adopts Participant Game and Perfect Bayesian Equilibrium to exploratively perform some aggressive actions to obtain additional information, thus the autonomous vehicle can cooperate robustly and efficiently. Explorative Game assumes each vehicle maintains a unique belief about the current situation and attributes insecurity and instability to the conflict of various Perfect Bayesian Equilibriums formed by various beliefs. Aggressive actions enable the ego car to proactively guide social vehicles to cooperate as it expects and encourage them to express their intentions as quickly and clearly as possible so that the equilibriums can converge and the conflict can be eliminated. Additional information reduces the error between the actual intentions of social vehicles and the estimated intentions from the ego car, helping rationally prune potential interactions and update parameters of the reward function. We demonstrate our algorithm on recorded data as well as virtual environments with manually controlled social vehicles to prove the efficiency of cooperation and the robustness of decision-making. And it has been running for more than 20 kilometers in the real world.

Imitation learning (IL) can generate computationally efficient sensorimotor policies from demonstrations provided by computationally expensive model-based sensing and control algorithms. However, commonly employed IL methods are often data-inefficient, requiring the collection of a large number of demonstrations and producing policies with limited robustness to uncertainties. In this work, we combine IL with an output feedback robust tube model predictive controller (RTMPC) to co-generate demonstrations and a data augmentation strategy to efficiently learn neural network-based sensorimotor policies. Thanks to the augmented data, we reduce the computation time and the number of demonstrations needed by IL, while providing robustness to sensing and process uncertainty. We tailor our approach to the task of learning a trajectory tracking visuomotor policy for an aerial robot, leveraging a 3D mesh of the environment as part of the data augmentation process. We numerically demonstrate that our method can learn a robust visuomotor policy from a single demonstration—a two-orders of magnitude improvement in demonstration efficiency compared to existing IL methods.

For robot and drone auditions, microphone arrays have been used for estimating sound source directions and sound source locations. By using sound source localization techniques, for example, drones can detect people calling for help even if the target person is not visible. Most sound source localization methods are based on estimated sound source directions and triangulation. However, when it comes to situations using drones, severe drone noise distorts direction estimation results which could worsen the localization results badly due to the discreteness of direction estimation. In this perspective, the authors have proposed a sound source localization method that can omit outlying triangulation points, which could improve its localization performance. In this paper, an outdoor experiment has been held, and the proposed method is evaluated whether it can localize a sound source even if real drone noise is added to the recordings. Experiment results show that the proposed method can localize with 4.15 m of estimation error for a sound source up to 50 m away, suppress the impact of outliers, and use only plausible triangulation points.

Inspired by the traditional art of paper folding, origami, autonomous production of 3D structures from 2D sheets can be achieved by the implementation of self-folding techniques. One technique to achieve such transformation is the usage of thermo-responsive smart materials such as self-folding polymeric films, which can be controlled by heat to shrink. Achieving remote self-folding with a practical approach remains a major challenge due to the requirement for specific environments, or having to accompany electronics on origami, which limits the complexity of the origami design. In this paper, we present a wireless method to trigger the thermo-responsive self-folding process of the origami robots through magnetic induction. The proposed method is applicable for all electrically conductive materials and can wirelessly fold a mobile origami robot with a size of 32 × 30 mm2. This method eliminates the need for inclusion of electronics on the origami or usage of complicated trigger methods and environmental conditions, allowing the robot to fold in a wider range of applications such as in constrained spaces.

Over the last decade, the ability to teach actions to robots in a user-friendly way has gained relevance, and a practical way of teaching robots a new task is to use Inverse Reinforcement Learning (IRL). In IRL, an expert teacher shows the robot a desired behaviour and an agent builds a model of the reward. The agent can also infer a policy that performs in an optimal way within the limitations of the knowledge provided to it. However, most IRL approaches assume an (almost) optimal performance of the teaching agent, which might become unpractical if the teacher is not actually an expert. In addition, most IRL focus on discrete state-action spaces that limit their applicability to certain real-world problems such as within the context of direct Policy Search (PS) reinforcement learning. Therefore, in this paper we introduce Ordinal Inverse Reinforcement Learning (OrdIRL) for continuous state variables, in which the teacher can qualitatively evaluate robot performance by selecting one among the predefined performance levels (e.g. {bad, medium, good} for three tiers of performance). Once the OrdIRL has fit an ordinal distribution to the data, we propose to use Bayesian Optimization (BO) to either gain knowledge on the inferred model (exploration) or find a policy or action that maximizes the expected reward given the prior knowledge on the reward (exploitation). In the case of large-dimensional state-action spaces, we use Dimensionality Reduction (DR) techniques and perform the BO in the latent space. Experimental results on simulation and with a robot arm show how this approach allows for learning the reward function with small data.

For performing robotic manipulation tasks, the core problem is determining suitable trajectories that fulfill the task requirements. Various approaches to compute such trajectories exist, being learning and optimization the main driving techniques. Our work builds on the learning-from-demonstration (LfD) paradigm, where an expert demonstrates motions, and the robot learns to imitate them. However, expert demonstrations are not sufficient to capture all sorts of task specifications, such as the timing to grasp an object. In this paper, we propose a new method that considers formal task specifications within LfD skills. Precisely, we leverage Signal Temporal Logic (STL), an expressive form of temporal properties of systems, to formulate task specifications and use black-box optimization (BBO) to adapt an LfD skill accordingly. We demonstrate our approach in simulation and on a real industrial setting using several tasks that showcase how our approach addresses the LfD limitations using STL and BBO.

How to design an optimal wearable device for human movement recognition is vital to reliable and accurate human-machine collaboration. Previous works mainly fabricate wearable devices heuristically. Instead, this paper raises an academic question: can we design an optimization algorithm to optimize the fabrication of wearable devices such as figuring out the best sensor arrangement automatically? Specifically, this work focuses on optimizing the placement of Forcemyography (FMG) sensors for FMG armbands in the application of arm movement recognition. Firstly, based on graph theory, the arm-band is modeled considering sensors' signals and connectivity. Then, a Graph-based Armband Modeling Network (GAM-Net) is introduced for arm movement recognition. Afterward, the sensor placement optimization for FMG armbands is formu-lated and an optimization algorithm with greedy local search is proposed. To study the effectiveness of our optimization algorithm, a dataset for mechanical maintenance tasks using FMG armbands with 16 sensors is collected. Our experiments show that using only 4 sensors optimized with our algorithm can help maintain a comparable recognition accuracy to using all sensors. Finally, the optimized sensor placement result is verified from a physiological view. This work would like to shed light on the automatic fabrication of wearable devices considering downstream tasks, such as human biological signal collection and movement recognition.

The reliability of autonomous traverses of rovers is critical. It may be jeopardized by the accumulation of errors and the uncertainty propagation of their localization systems. Moreover, space environments are usually harsh, challenging and unpredictable. Teleoperation is complex due to the significant and unavoidable delay. For these reasons, a path planner that provides some level of autonomy with guarantees could increase the success rate of planetary exploration missions. This paper proposes a path planning solution that tackles increasing localization uncertainty and makes a trade-off between the collision risk and the path length. The planner uses the the Fast Marching Method (FMM) to produce a costmap aware of this uncertainty and calculate the optimal path for a level of confidence. This paper additionally presents several simulation and experimental using a wheeled robotic vehicle within a lunar analogue facility.

Most deformable object manipulation tasks still rely on skillful human operators. To automate such tasks, a robotic system should not only be able to deform an object to a desired shape but also servo its deformation along a specific path towards the desired shape. We propose a shape servoing control scheme to automate such tasks. Our scheme controls the deformation trajectory towards the desired shape by imposing task-focused convergence constraints. The constraints impose how fast the different regions of the object converge to the desired shape. Integrating such a behavior in shape servoing forms our main contribution. Experiments, carried out on rubber layer assembly tasks, show that our control scheme outperforms a state-of-the-art shape servoing scheme.

Intercepting an object in flight through nonpre-hensile manipulation is a challenging problem, which is aimed at catching and stopping a flying object using little contacts without completely restraining its relative motion to the robot. This paper presents a two-stage optimal trajectory generation method to tackle this problem. At the pre-catching stage, optimal position and attitude trajectories of the robot's end-effector to approach the object are generated by a variational method. At the post-catching stage, the end-effector's trajectories are generated to optimally eliminate the translational and rotational motion of the object and a convex-MPC algorithm combined with admittance control is used to realize the trajectory tracking. A series of simulations and experiments have been conducted to verify the effectiveness of the proposed method.

In multi-robot missions, relative position and attitude information between robots is valuable for a variety of tasks such as mapping, planning, and formation control. In this paper, the problem of estimating relative poses from a set of inter-robot range measurements is investigated. Specifically, it is shown that the estimation accuracy is highly dependent on the true relative poses themselves, which prompts the desire to find multi-robot formations that provide the best estimation performance. By direct maximization of Fischer information, it is shown in simulation and experiment that large improvements in estimation accuracy can be obtained by optimizing the formation geometry of a team of robots.

Position estimation in Multi-Robot Systems (MRS) relies on relative angle or distance measurements between the robots, which generally deteriorate as distances increase. Moreover, the localization accuracy is strongly influenced both by the quality of the raw measurements but also by the overall geometry of the network. In this paper, we design a cost function that accounts for these two issues and can be used to develop motion planning algorithms that optimize the localizability in MRS, i.e., the ability of individual robots to localize themselves accurately. This cost function is based on computing new Cramér Rao Lower Bounds characterizing the achievable positioning performance with range and angle measurements that deteriorate with increasing distances. We describe a gradient-based motion-planning algorithm for MRS deployment that can be implemented in a distributed manner, as well as a non-myopic strategy to escape local minima. Finally, we test the proposed methodology experimentally for range measurements obtained using ultra-wide band transceivers and illustrate the improvements resulting from leveraging the more accurate measurement model in the robot placement algorithms.