ICRA 2024 论文列表

Despite large advances in recent years, real-time capable motion planning for autonomous road vehicles remains a huge challenge. In this work, we present a decision module that is based on set-based reachability analysis: First, we identify all possible driving corridors by computing the reachable set for the longitudinal position of the vehicle along the lanelets of the road network, where lane changes are modeled as discrete events. Next, we select the best driving corridor based on a cost function that penalizes lane changes and deviations from a desired velocity profile. Finally, we generate a reference trajectory inside the selected driving corridor, which can be used to guide or warm start low-level trajectory planners. For the numerical evaluation we combine our decision module with a motion-primitive-based and an optimization-based planner and evaluate the performance on 2000 challenging CommonRoad traffic scenarios as well in the realistic CARLA simulator. The results demonstrate that our decision module is real-time capable and yields significant speed-ups compared to executing a motion planner standalone without a decision module.

This paper presents an adaptive online learning framework for systems with uncertain parameters to ensure safety-critical control in non-stationary environments. Our approach consists of two phases. The initial phase is centered on a novel sparse Gaussian process (GP) framework. We first integrate a forgetting factor to refine a variational sparse GP algorithm, thus enhancing its adaptability. Subsequently, the hyperparameters of the Gaussian model are trained with a specially compound kernel, and the Gaussian model’s online inferential capability and computational efficiency are strengthened by updating a solitary inducing point derived from newly samples, in conjunction with the learned hyperparameters. In the second phase, we propose a safety filter based on high order control barrier functions (HOCBFs), synergized with the previously trained learning model. By leveraging the compound kernel from the first phase, we effectively address the inherent limitations of GPs in handling high-dimensional problems for real-time applications. The derived controller ensures a rigorous lower bound on the probability of satisfying the safety specification. Finally, the efficacy of our proposed algorithm is demonstrated through real-time obstacle avoidance experiments executed using both simulation platform and a real-world 7-DOF robot.

We present ReLU-QP, a GPU-accelerated solver for quadratic programs (QPs) that is capable of solving high-dimensional control problems at real-time rates. ReLU-QP is derived by exactly reformulating the Alternating Direction Method of Multipliers (ADMM) algorithm for solving QPs as a deep, weight-tied neural network with rectified linear unit (ReLU) activations. This reformulation enables the deployment of ReLU-QP on GPUs using standard machine-learning toolboxes. We evaluate the performance of ReLU-QP across three model-predictive control (MPC) benchmarks: stabilizing random linear dynamical systems with control limits, balancing an Atlas humanoid robot on a single foot, and performing a whole-body pick-up motion on a quadruped equipped with a six-degree-of-freedom arm. These benchmarks indicate that ReLU-QP is competitive with state-of-the-art CPU-based solvers for small-to-medium-scale problems and offers order-of-magnitude speed improvements for larger-scale problems.

This paper presents an aerial robotic platform for rapid remote elevated overhead-perching drill operations for wood health inspection. The platform features an innovative passive prismatic-gripper mechanism affixed to the aerial robot’s top, facilitating overhead drilling. The primary aim is to enhance the safety and efficiency of elevated wood structure inspection using the resistography method, which involves drilling into wooden structures to identify internal voids. The research centers on two key enabling technologies: a gripper mechanism for secure attachment to target surfaces and a tethered drill configuration for drilling operations. The novel gripper mechanism enables drilling on large planar surfaces and even small beam-width structures. The paper concludes with discussions on design simulations and drill resistance experiments, highlighting the effectiveness of the proposed approach in detecting internal cavities within wooden structures.

Teaching robots novel skills with demonstrations via human-in-the-loop data collection techniques like kinesthetic teaching or teleoperation puts a heavy burden on human supervisors. In contrast to this paradigm, it is often significantly easier to provide raw, action-free visual data of tasks being performed. Moreover, this data can even be mined from video datasets or the web. Ideally, this data can serve to guide robot learning for new tasks in novel environments, informing both "what" to do and "how" to do it. A powerful way to encode both the "what" and the "how" is to infer a well-shaped reward function for reinforcement learning. The challenge is determining how to ground visual demonstration inputs into a well-shaped and informative reward function. We propose a technique Rank2Reward for learning behaviors from videos of tasks being performed without access to any low-level states and actions. We do so by leveraging the videos to learn a reward function that measures incremental "progress" through a task by learning how to temporally rank the video frames in a demonstration. By inferring an appropriate ranking, the reward function is able to guide reinforcement learning by indicating when task progress is being made. This ranking function can be integrated into an adversarial imitation learning scheme resulting in an algorithm that can learn behaviors without exploiting the learned reward function. We demonstrate the effectiveness of Rank2Reward at learning behaviors from raw video on a number of tabletop manipulation tasks in both simulations and on a real-world robotic arm. We also demonstrate how Rank2Reward can be easily extended to be applicable to web-scale video datasets. Code and videos are available at https://rank2reward.github.io

Camera-based tactile sensors can provide high resolution positional and local geometry information for robotic manipulation. Curved and rounded fingers are often advantageous, but it can be difficult to derive illumination systems that work well within curved geometries. To address this issue, we introduce RainbowSight, a family of curved, compact, camera-based tactile sensors which use addressable RGB LEDs illuminated in a novel rainbow spectrum pattern. In addition to being able to scale the illumination scheme to different sensor sizes and shapes to fit on a variety of end effector configurations, the sensors can be easily manufactured and require minimal optical tuning to obtain high resolution depth reconstructions of an object deforming the sensor’s soft elastomer surface. Additionally, we show the advantages of our new hardware design and improvements in calibration methods for accurate depth map generation when compared to alternative lighting methods commonly implemented in previous camera-based tactile sensors. With these advancements, we make the integration of tactile sensors more accessible to roboticists by allowing them the flexibility to easily customize, fabricate, and calibrate camera-based tactile sensors to best fit the needs of their robotic systems.

We present a novel approach for metric dense depth estimation based on the fusion of a single-view image and a sparse, noisy Radar point cloud. The direct fusion of heterogeneous Radar and image data, or their encodings, tends to yield dense depth maps with significant artifacts, blurred boundaries, and suboptimal accuracy. To circumvent this issue, we learn to augment versatile and robust monocular depth prediction with the dense metric scale induced from sparse and noisy Radar data. We propose a Radar-Camera framework for highly accurate and fine-detailed dense depth estimation with four stages, including monocular depth prediction, global scale alignment of monocular depth with sparse Radar points, quasi-dense scale estimation through learning the association between Radar points and image patches, and local scale refinement of dense depth using a scale map learner. Our proposed method significantly outperforms the state-of-the-art Radar-Camera depth estimation methods by reducing the mean absolute error (MAE) of depth estimation by 25.6% and 40.2% on the challenging nuScenes dataset and our self-collected ZJU-4DRadarCam dataset, respectively. Our code and dataset will be released at https://github.com/MMOCKING/RadarCam-Depth.

Odometry and mapping play a pivotal role in the navigation of autonomous vehicles. In this paper, we address the problem of pose estimation and map creation using only radar sensors. We focus on two odometry estimation approaches followed by a mapping step. The first one is a new point-to-point ICP approach that leverages the velocity information provided by 3D radar sensors. The second one is advantageous for 2D radars with a low number of samples, and particularly useful for scenarios where the sensor is being blocked by large dynamic obstacles. It exploits a constant velocity filter and the measured Doppler velocities to estimate the vehicle’s ego-motion. We enrich this with a filtering step to improve the accuracy of the points in the resulting map. We put our work to the test using the View of Delft and NuScenes datasets, which involve 3D and 2D radar sensors. Our findings illustrate state-of-the-art performance of our odometry techniques in terms of accuracy when compared to existing alternatives. Moreover, we demonstrate that our map filtering methodology achieves higher similarity rates than the raw unfiltered map when benchmarked against a corresponding LiDAR map.

Robots and autonomous vehicles should be aware of what happens in their surroundings. The segmentation and tracking of moving objects are essential for reliable path planning, including collision avoidance. We investigate this estimation task for vehicles using radar sensing. We address moving instance tracking in sparse radar point clouds to enhance scene interpretation. We propose a learning-based radar tracker incorporating temporal offset predictions to enable direct center-based association and enhance segmentation performance by including additional motion cues. We implement attention-based tracking for sparse radar scans to include appearance features and enhance performance. The final association combines geometric and appearance features to overcome the limitations of center-based tracking to associate instances reliably. Our approach shows an improved performance on the moving instance tracking benchmark of the RadarScenes dataset compared to the current state of the art.

In this work, we present RadCloud, a novel real-time framework for directly obtaining higher-resolution lidar-like 2D point clouds from low-resolution radar frames on resource-constrained platforms commonly used in unmanned aerial and ground vehicles (UAVs and UGVs, respectively); such point clouds can then be used for accurate environmental mapping, navigating unknown environments, and other robotics tasks. While high-resolution sensing using radar data has been previously reported, existing methods cannot be used on most UAVs, which have limited computational power and energy; thus, existing demonstrations focus on offline radar processing. RadCloud overcomes these challenges by using a radar configuration with 1/4th of the range resolution and employing a deep learning model with 2.25× fewer parameters. Additionally, RadCloud utilizes a novel chirp-based approach that makes obtained point clouds resilient to rapid movements (e.g., aggressive turns or spins) that commonly occur during UAV flights. In real-world experiments, we demonstrate the accuracy and applicability of RadCloud on commercially available UAVs and UGVs, with off-the-shelf radar platforms on-board.

Mobile autonomy relies on the precise perception of dynamic environments. Robustly tracking moving objects in 3D world thus plays a pivotal role for applications like trajectory prediction, obstacle avoidance, and path planning. While most current methods utilize LiDARs or cameras for Multiple Object Tracking (MOT), the capabilities of 4D imaging radars remain largely unexplored. Recognizing the challenges posed by radar noise and point sparsity in 4D radar data, we introduce RaTrack, an innovative solution tailored for radar-based tracking. Bypassing the typical reliance on specific object types and 3D bounding boxes, our method focuses on motion segmentation and clustering, enriched by a motion estimation module. Evaluated on the View-of-Delft dataset, RaTrack showcases superior tracking precision of moving objects, largely surpassing the performance of the state of the art. We release our code and model at https://github.com/LJacksonPan/RaTrack.

In robotic vision, a de-facto paradigm is to learn in simulated environments and then transfer to real-world applications, which poses an essential challenge in bridging the sim-to-real domain gap. While mainstream works tackle this problem in the RGB domain, we focus on depth data synthesis and develop a Range-aware RGB-D data Simulation pipeline (RaSim). In particular, high-fidelity depth data is generated by imitating the imaging principle of real-world sensors. A range-aware rendering strategy is further introduced to enrich data diversity. Extensive experiments show that models trained with RaSim can be directly applied to real-world scenarios without any finetuning and excel at downstream RGB-D perception tasks. Data and code are available at https://github.com/shanice-l/RaSim.

Localization is paramount for autonomous robots. While camera and LiDAR-based approaches have been extensively investigated, they are affected by adverse illumination and weather conditions. Therefore, radar sensors have recently gained attention due to their intrinsic robustness to such conditions. In this paper, we propose RaLF, a novel deep neural network-based approach for localizing radar scans in a LiDAR map of the environment, by jointly learning to address both place recognition and metric localization. RaLF is composed of radar and LiDAR feature encoders, a place recognition head that generates global descriptors, and a metric localization head that predicts the 3-DoF transformation between the radar scan and the map. We tackle the place recognition task by learning a shared embedding space between the two modalities via cross-modal metric learning. Additionally, we perform metric localization by predicting pixel-level flow vectors that align the query radar scan with the LiDAR map. We extensively evaluate our approach on multiple real-world driving datasets and show that RaLF achieves state-of-the-art performance for both place recognition and metric localization. Moreover, we demonstrate that our approach can effectively generalize to different cities and sensor setups than the ones used during training. We make the code and trained models publicly available at http://ralf.cs.uni-freiburg.de.

Robust underwater motion planning of autonomous underwater vehicles (AUVs) in dynamic cluttered environments is a problem that has yet to be addressed in depth. Due to advances in technology and computational capacity, AUVs are expected to operate safely and autonomously in increasingly challenging environments, necessitating methods that are able to safely navigate robots in real-time. Though, most solutions remain overly cautious and conservative. This paper proposes RUMP, a novel locally-optimal motion planning framework for robust real-time autonomous underwater navigation in 3D cluttered environments consisting of observed static and dynamic obstacles. The problem is modeled using path optimization and can be solved in real-time with a common nonlinear solver. The constructed objective function allows deciding the local goal during optimization to both maximize safety within a planning horizon and minimize the expected distance to the target position. Furthermore, path safety is considered for the entire transition between consecutive states, utilizing a novel approach for continuous spatiotemporal collision checks. The proposed formulation provides safe performance even in environments with obstacles that may move orders of magnitude faster than the AUV itself. Simulation experiments, in different challenging scenarios of obstacles moving up to 100 times faster than the robot, showcase robustness and efficient real-time performance of more than 15 Hz.

Numerous datasets and benchmarks exist to assess and compare Simultaneous Localization and Mapping (SLAM) algorithms. Nevertheless, their precision must follow the rate at which SLAM algorithms improved in recent years. Moreover, current datasets fall short of comprehensive data-collection protocol for reproducibility and the evaluation of the precision or accuracy of the recorded trajectories. With this objective in mind, we proposed the Robotic Total Stations Ground Truthing dataset (RTS-GT) dataset to support localization research with the generation of six-Degrees Of Freedom (DOF) ground truth trajectories. This novel dataset includes six-DOF ground truth trajectories generated using a system of three Robotic Total Stations (RTSs) tracking moving robotic platforms. Furthermore, we compare the performance of the RTS-based system to a Global Navigation Satellite System (GNSS)-based setup. The dataset comprises around sixty experiments conducted in various conditions over a period of 17 months, and encompasses over 49 kilometers of trajectories, making it the most extensive dataset of RTS-based measurements to date. Additionally, we provide the precision of all poses for each experiment, a feature not found in the current state-of-the-art datasets. Our results demonstrate that RTSs provide measurements that are 22 times more stable than GNSS in various environmental settings, making them a valuable resource for SLAM benchmark development.

In order to successfully perform manipulation tasks in new environments, such as grasping, robots must be proficient in segmenting unseen objects from the background and/or other objects. Previous works perform unseen object instance segmentation (UOIS) by training deep neural networks on large-scale data to learn RGB/RGB-D feature embeddings, where cluttered environments often result in inaccurate segmentations. We build upon these methods and introduce a novel approach to correct inaccurate segmentation, such as under-segmentation, of static image-based UOIS masks by using robot interaction and a designed body frame-invariant feature. We demonstrate that the relative linear and rotational velocities of frames randomly attached to rigid bodies due to robot interactions can be used to identify objects and accumulate corrected object-level segmentation masks. By introducing motion to regions of segmentation uncertainty, we are able to drastically improve segmentation accuracy in an uncertainty-driven manner with minimal, non-disruptive interactions (ca. 2-3 per scene). We demonstrate the effectiveness of our proposed interactive perception pipeline in accurately segmenting cluttered scenes by achieving an average object segmentation accuracy rate of 80.7%, an increase of 28.2% when compared with other state-of-the-art UOIS methods.

Autonomous navigation in unstructured environments is a challenging task due to the complex and dynamic nature of robot-terrain interactions. Existing approaches often struggle to generalize amidst the complexities of real-world settings. They tend to rely on hand-engineered, rule-based robot models or static weightings assigned to obstacles, semantics, and other perceptual cues to estimate traversability. To address these challenges, we propose a novel approach called Reinforcement-Based Inferred Dynamics via Emulating Rehearsals (RIDER), that learns the dynamics of robot-terrain interactions within a compact latent space, capturing robot’s traversability. Operating within a reinforcement learning paradigm, RIDER learns to infer its own dynamics by predicting how future robot observations and states evolve within this latent space in response to navigational behaviors. Furthermore, our approach leverages emulated rehearsals, where the robot learns within the latent space to predict its rewards and generate navigational behaviors, even when real observations have not been updated. Accordingly, RIDER equips robots with the ability to generate navigational behaviors by predicting environmental changes, and plan beyond the speed at which observations from sensors are available. Experimental results and comparisons with baseline methods establish that our proposed method outperforms other approaches in cluttered and unstructured environments and demonstrates an enhanced capacity for autonomous navigation in real-world settings.

Unlike in natural images, in endoscopy there is no clear notion of an up-right camera orientation. Endoscopic videos therefore often contain large rotational motions, which require keypoint detection and description algorithms to be robust to these conditions. While most classical methods achieve rotation-equivariant detection and invariant description by design, many learning-based approaches learn to be robust only up to a certain degree. At the same time learning-based methods under moderate rotations often outperform classical approaches. In order to address this shortcoming, in this paper we propose RIDE, a learning-based method for rotation-equivariant detection and invariant description. Following recent advancements in group-equivariant learning, RIDE models rotation-equivariance implicitly within its architecture. Trained in a self-supervised manner on a large curation of endoscopic images, RIDE requires no manual labeling of training data. We test RIDE in the context of surgical tissue tracking on the SuPeR dataset as well as in the context of relative pose estimation on a repurposed version of the SCARED dataset. In addition we perform explicit studies showing its robustness to large rotations. Our comparison against recent learning-based and classical approaches shows that RIDE sets a new state-ofthe-art performance on matching and relative pose estimation tasks and scores competitively on surgical tissue tracking.

General scene reconstruction refers to the task of estimating the full 3D geometry and texture of a scene containing previously unseen objects. In many practical applications such as AR/VR, autonomous navigation, and robotics, only a single view of the scene may be available, making the scene reconstruction task challenging. In this paper, we present a method for scene reconstruction by structurally breaking the problem into two steps: rendering novel views via inpainting and 2D to 3D scene lifting. Specifically, we leverage the generalization capability of large visual language models (DALL•E 2) to inpaint the missing areas of scene color images rendered from different views. Next, we lift these inpainted images to 3D by predicting normals of the inpainted image and solving for the missing depth values. By predicting for normals instead of depth directly, our method allows for robustness to changes in depth distributions and scale. With rigorous quantitative evaluation, we show that our method outperforms multiple baselines while providing generalization to novel objects and scenes. Code and data is available at https://samsunglabs.github.io/RIC-project-page/.