ICRA 2024 论文列表

We propose a topological mapping and localization system able to operate on real human colonoscopies, despite significant shape and illumination changes. The map is a graph where each node codes a colon location by a set of real images, while edges represent traversability between nodes. For close-in-time images, where scene changes are minor, place recognition can be successfully managed with the recent transformers-based local feature matching algorithms. However, under long-term changes –such as different colonoscopies of the same patient– feature-based matching fails. To address this, we train on real colonoscopies a deep global descriptor achieving high recall with significant changes in the scene. The addition of a Bayesian filter boosts the accuracy of long-term place recognition, enabling relocalization in a previously built map. Our experiments show that ColonMapper is able to autonomously build a map and localize against it in two important use cases: localization within the same colonoscopy or within different colonoscopies of the same patient. Code: github.com/jmorlana/ColonMapper.

State-of-the-art techniques for monocular camera reconstruction predominantly rely on the Structure from Motion (SfM) pipeline. However, such methods often yield reconstruction outcomes that lack crucial scale information, and over time, accumulation of images leads to inevitable drift issues. In contrast, mapping methods based on LiDAR scans are popular in large-scale urban scene reconstruction due to their precise distance measurements, a capability fundamentally absent in visual-based approaches. Researchers have made attempts to utilize concurrent LiDAR and camera measurements in pursuit of precise scaling and color details within mapping outcomes. However, the outcomes are subject to extrinsic calibration and time synchronization precision. In this paper, we propose a novel cost-effective reconstruction pipeline that utilizes a pre-established LiDAR map as a fixed constraint to effectively address the inherent scale challenges present in monocular camera reconstruction. To our knowledge, our method is the first to register images onto the point cloud map without requiring synchronous capture of camera and LiDAR data, granting us the flexibility to manage reconstruction detail levels across various areas of interest. To facilitate further research in this domain, we have released Colmap-PCD3, an open-source tool leveraging the Colmap algorithm, that enables precise fine-scale registration of images to the point cloud map.

To date, commercial industrial robots only provide multi-robot coordination for their own fleet of robots and treat robots from other vendors as general obstacles. The ability to enable robots from different vendors to co-exist in the same space is crucial to prevent vendor lock-in. We present the first decentralized system that achieves coordination between a heterogeneous fleet of black box robots for which the internals of the navigation stack are presumed unmodifiable. Our system, which we call CODAK, achieves the coordination by relying on minimum set of interfaces that are commonly available on most industrial and service robots. For each robot, CODAK uses a trained recurrent neural network to anticipate collisions from externally observable metrics. Anticipated collisions are avoided using a simple, but yet effective, concurrency control scheme. We run a series of experiments in simulation and with real robots to demonstrate CODAK’s ability to enable safe navigation in different environments. We also experimentally compare CODAK with previously published white-box solutions to evaluate the penalty of black-box constraint.

End-to-end deep reinforcement learning (DRL) for quadrotor control promises many benefits – easy deployment, task generalization and real-time execution capability. Prior end-to-end DRL-based methods have showcased the ability to deploy learned controllers onto single quadrotors or quadrotor teams maneuvering in simple, obstacle-free environments. However, the addition of obstacles increases the number of possible interactions exponentially, thereby increasing the difficulty of training RL policies. In this work, we propose an end-to-end DRL approach to control quadrotor swarms in environments with obstacles. We provide our agents a curriculum and a replay buffer of the clipped collision episodes to improve performance in obstacle-rich environments. We implement an attention mechanism to attend to the neighbor robots and obstacle interactions - the first successful demonstration of this mechanism on policies for swarm behavior deployed on severely compute-constrained hardware. Our work is the first work that demonstrates the possibility of learning neighbor-avoiding and obstacle-avoiding control policies trained with end-to-end DRL that transfers zero-shot to real quadrotors. Our approach scales to 32 robots with 80% obstacle density in simulation and 8 robots with 20% obstacle density in physical deployment. Website: https://sites.google.com/view/obst-avoid-swarm-rl

Manipulating deformable objects arises in daily life and numerous applications. Despite phenomenal advances in industrial robotics, manipulation of deformable objects remains mostly a manual task. This is because of the high number of internal degrees of freedom and the complexity of predicting its motion. In this paper, we apply the computationally efficient position-based dynamics method to predict object motion and distance to obstacles. This distance is incorporated in a control barrier function for the resolved motion kinematic control for one or more robots to adjust their motion to avoid colliding with the obstacles. The controller has been applied in simulations to 1D and 2D deformable objects with varying numbers of assistant agents, demonstrating its versatility across different object types and multi-agent systems. Results indicate the feasibility of real-time collision avoidance through deformable object simulation, minimizing path tracking error while maintaining a predefined minimum distance from obstacles and preventing overstretching of the deformable object. The implementation is performed in ROS, allowing ready portability to different applications.

Maps have played an indispensable role in enabling safe and automated driving. Although there have been many advances on different fronts ranging from SLAM to semantics, building an actionable hierarchical semantic representation of urban dynamic scenes and processing information from multiple agents are still challenging problems. In this work, we present Collaborative URBan Scene Graphs (CURB-SG) that enable higher-order reasoning and efficient querying for many functions of automated driving. CURB-SG leverages panoptic LiDAR data from multiple agents to build large-scale maps using an effective graph-based collaborative SLAM approach that detects inter-agent loop closures. To semantically decompose the obtained 3D map, we build a lane graph from the paths of ego agents and their panoptic observations of other vehicles. Based on the connectivity of the lane graph, we segregate the environment into intersecting and non-intersecting road areas. Subsequently, we construct a multi-layered scene graph that includes lane information, the position of static landmarks and their assignment to certain map sections, other vehicles observed by the ego agents, and the pose graph from SLAM including 3D panoptic point clouds. We extensively evaluate CURB-SG in urban scenarios using a photorealistic simulator. We release our code at http://curb.cs.uni-freiburg.de.

Collaborative decision-making is an essential capability for multi-robot systems, such as connected vehicles, to collaboratively control autonomous vehicles in accident-prone scenarios. Under limited communication bandwidth, capturing comprehensive situational awareness by integrating connected agents’ observation is very challenging. In this paper, we propose a novel collaborative decision-making method that efficiently and effectively integrates collaborators’ representations to control the ego vehicle in accident-prone scenarios. Our approach formulates collaborative decision-making as a classification problem. We first represent sequences of raw observations as spatiotemporal graphs, which significantly reduce the package size to share among connected vehicles. Then we design a novel spatiotemporal graph neural network based on heterogeneous graph learning, which analyzes spatial and temporal connections of objects in a unified way for collaborative decision-making. We evaluate our approach using a high-fidelity simulator that considers realistic traffic, communication bandwidth, and vehicle sensing among connected autonomous vehicles. The experimental results show that our representation achieves over 100x reduction in the shared data size that meets the requirements of communication bandwidth for connected autonomous driving. In addition, our approach achieves over 30% improvements in driving safety.

Retinal surgery is a complex medical procedure that requires high precision dexterity to perform delicate instrument maneuvers with sub-millimeter accuracy. Minimizing the manual tremor and achieving precise and repeatable execution of surgical tasks has motivated the development of robotic platforms to overcome the limitations of manual surgery. However, specific tasks, such as instrument insertion through the trocar, are more challenging in robotic surgery than in conventional manual procedures since the robot control is often optimized for navigation inside the eye. This challenges the integration of robotic systems, creating a high cognitive load on the operator and prolonging the surgery time. Moreover, misalignment of the robot’s remote center of motion (RCM) and trocar position during the procedure can lead to excessive forces between the instrument and the trocar, potentially causing patient trauma. Precise and rapid localization of the trocars enables the automation of the insertion procedure and dynamic compensation of eye motion.In this work, we present a real-time marker-less method for 3D pose tracking of trocar, achieved with only a single monocular camera. Our experiments show promising results towards real-time trocar pose estimation and tracking, achieving an average error of 3◦ in trocar orientation estimation, with an average processing time of 15 fps. This could serve as a foundation to improve robotic systems’ automation, integration, and efficiency of robotic systems for retinal surgery. The dataset created for this work is made publicly available.

Perception is necessary for autonomous navigation in an unknown area crowded with obstacles. It’s challenging for a robot to navigate safely without any sensors that can sense the environment, resulting in a blind robot, and becomes more difficult when comes to a group of robots. However, it could be costly to equip all robots with expensive perception or SLAM systems. In this paper, we propose a novel system named ColAG, to solve the problem of autonomous navigation for a group of blind UGVs by introducing cooperation with one UAV, which is the only robot that has full perception capabilities in the group. The UAV uses SLAM for its odometry and mapping while sharing this information with UGVs via limited relative pose estimation. The UGVs plan their trajectories in the received map and predict possible failures caused by the uncertainty of its wheel odometry and unknown risky areas. The UAV dynamically schedules waypoints to prevent UGVs from collisions, formulated as a Vehicle Routing Problem with Time Windows to optimize the UAV’s trajectories and minimize time when UGVs have to wait to guarantee safety. We validate our system through extensive simulation with up to 7 UGVs and real-world experiments with 3 UGVs.

In the pursuit of fully autonomous robotic systems capable of taking over tasks traditionally performed by humans, the complexity of open-world environments poses a considerable challenge. Addressing this imperative, this study contributes to the field of Large Language Models (LLMs) applied to task and motion planning for robots. We propose a system architecture that orchestrates a seamless interplay between multiple cognitive levels, encompassing reasoning, planning, and motion generation. At its core lies a novel replanning strategy that handles physically grounded, logical, and semantic errors in the generated plans. We demonstrate the efficacy of the proposed feedback architecture, particularly its impact on executability, correctness, and time complexity via empirical evaluation in the context of a simulation and two intricate real-world scenarios: blocks world, barman and pizza preparation.

Collaborative state estimation using different heterogeneous sensors is a fundamental prerequisite for robotic swarms operating in GPS-denied environments, posing a significant research challenge. In this paper, we introduce a centralized system to facilitate collaborative LiDAR-ranging-inertial state estimation, enabling robotic swarms to operate without the need for anchor deployment. The system efficiently distributes computationally intensive tasks to a central server, thereby reducing the computational burden on individual robots for local odometry calculations. The server back-end establishes a global reference by leveraging shared data and refining joint pose graph optimization through place recognition, global optimization techniques, and removal of outlier data to ensure precise and robust collaborative state estimation. Extensive evaluations of our system, utilizing both publicly available datasets and our custom datasets, demonstrate significant enhancements in the accuracy of collaborative SLAM estimates. Moreover, our system exhibits remarkable proficiency in large-scale missions, seamlessly enabling ten robots to collaborate effectively in performing SLAM tasks. In order to contribute to the research community, we will make our code open-source and accessible at https://github.com/PengYu-team/Co-LRIO.

Prior robot painting and drawing work, such as FRIDA, has focused on decreasing the sim-to-real gap and expanding input modalities for users, but the interaction with these systems generally exists only in the input stages. To support interactive, human-robot collaborative painting, we introduce the Collaborative FRIDA (CoFRIDA) robot painting framework, which can co-paint by modifying and engaging with content already painted by a human collaborator. To improve text-image alignment–FRIDA’s major weakness–our system uses pre-trained text-to-image models; however, pre-trained models in the context of real-world co-painting do not perform well because they (1) do not understand the constraints and abilities of the robot and (2) cannot perform co-painting without making unrealistic edits to the canvas and overwriting content. We propose a self-supervised fine-tuning procedure that can tackle both issues, allowing the use of pre-trained state-of-the-art text-image alignment models with robots to enable co-painting in the physical world. Our open-source approach, CoFRIDA, creates paintings and drawings that match the input text prompt more clearly than FRIDA, both from a blank canvas and one with human created work. More generally, our fine-tuning procedure successfully encodes the robot’s constraints and abilities into a foundation model, showcasing promising results as an effective method for reducing sim-to-real gaps. https://pschaldenbrand.github.io/cofrida/

Mass customization and shorter manufacturing cycles are becoming more important among small and medium-sized companies. However, classical industrial robots struggle to cope with product variation and dynamic environments. In this paper, we present CoBT, a collaborative programming by demonstration framework for generating reactive and modular behavior trees. CoBT relies on a single demonstration and a combination of data-driven machine learning methods with logic-based declarative learning to learn a task, thus eliminating the need for programming expertise or long development times. The proposed framework is experimentally validated on 7 manipulation tasks and we show that CoBT achieves ≈ 93% success rate overall with an average of 7.5s programming time. We conduct a pilot study with non-expert users to provide feedback regarding the usability of CoBT. More videos and generated behavior trees are available at: https://github.com/jainaayush2006/CoBT.git.

Selecting an optimal robot, its base pose, and trajectory for a given task is currently mainly done by human expertise or trial and error. To evaluate automatic approaches to this combined optimization problem, we introduce a benchmark suite encompassing a unified format for robots, environments, and task descriptions. Our benchmark suite is especially useful for modular robots, where the multitude of robots that can be assembled creates a host of additional parameters to optimize. We include tasks such as machine tending and welding in synthetic environments and 3D scans of real-world machine shops. All benchmarks are accessible through cobra.cps.cit.tum.de, a platform to conveniently share, reference, and compare tasks, robot models, and solutions.

Synthesizing planning and control policies in robotics is a fundamental task, further complicated by factors such as complex logic specifications and high-dimensional robot dynamics. This paper presents a novel reinforcement learning approach to solving high-dimensional robot navigation tasks with complex logic specifications by co-learning planning and control policies. Notably, this approach significantly reduces the sample complexity in training, allowing us to train high-quality policies with much fewer samples compared to existing reinforcement learning algorithms. In addition, our methodology streamlines complex specification extraction from map images and enables the efficient generation of long-horizon robot motion paths across different map layouts. Moreover, our approach also demonstrates capabilities for high-dimensional control and avoiding suboptimal policies via policy alignment. The efficacy of our approach is demonstrated through experiments involving simulated high-dimensional quadruped robot dynamics and a real-world differential drive robot (TurtleBot3) under different types of task specifications.

A robotic system’s hardware and control policy must be co-optimized to ensure they complement each other to interact robustly with the environment. However, this combined search is extremely high-dimensional and intractable without a suitable underlying representation. This paper uses environmental constraints to structure the co-design space for manipulation. We show that task-relevant constraints encode regions of the search space containing reasonable co-design solutions. Furthermore, this underlying representation renders a co-design space amenable to gradient-based optimization. For efficient search, we present the co-design Jacobian that describes how the robot’s motion varies with control as well as hardware design changes. This Jacobian exploits the structure induced by environmental constraints for iterative design updates in the co-design space. Using these two conceptual tools, we co-design manipulators, grippers, and multi-fingered hands, showing that environmental constraints are an effective representation for co-designing diverse manipulation systems. Our methodology also scales well with increased co-design parameters, rendering the co-design of complex, high-dimensional manipulation systems feasible.

The design and control of winged aircraft and drones is an iterative process aimed at identifying a compromise of mission-specific costs and constraints. When agility is required, shape-shifting (morphing) drones represent an efficient solution. However, morphing drones require the addition of actuated joints that increase the topology and control coupling, making the design process more complex. We propose a co-design optimisation method that assists the engineers by proposing a morphing drone’s conceptual design that includes topology, actuation, morphing strategy, and controller parameters. The method consists of applying multi-objective constraint-based optimisation to a multi-body winged drone with trajectory optimisation to solve the motion intelligence problem under diverse flight mission requirements, such as energy consumption and mission completion time. We show that co-designed morphing drones outperform fixed-winged drones in terms of energy efficiency and mission time, suggesting that the proposed co-design method could be a useful addition to the aircraft engineering toolbox.

Transorbital Neuro Surgery (TNS) offers a novel treatment towards the lesion inside skull pursuing minimal invasiveness. Most conventional TNS tools are rigid and straight, limiting the dexterity and accessibility in passing a small port. Bendable and steerable surgical tools provides an alternative for this issue. In this work, we proposed a dual-segment slender surgical robot arm for TNS, which is a Co-Axial Slender Tubular robot (CAST), and modelled it using novel approaches. Another contribution is tendon-mortise shaped slits along the axial direction, enhancing the overall stiffness. The bending of CAST is actuated by pushing/pulling distance, and the maximum diameter is only 1.7mm with high dexterity after mounting on a rigid robot arm. Experiments demonstrates that the proposed the slit design doubles the stiffness properties compared to traditional rectangle slit designs. The path-following task shows that the position error was maximally 3mm in open-looped control. Test on a skull model demonstrates that the whole system could successfully perform electrocoagulation procedure inside the depth of skull in a robotized manner effectively.

We present CloudGripper, an open source cloud robotics testbed, consisting of a scalable, space and cost- efficient design constructed as a rack of 32 small robot arm work cells. Each robot work cell is fully enclosed and features individual lighting, a low-cost Cartesian robot arm with an attached rotatable parallel jaw gripper and a dual camera setup for experimentation. The system design is focused on continuous operation and features a 10 Gbit/s network connectivity allowing for high throughput remote-controlled experimentation and data collection for robotic manipulation. Furthermore, CloudGripper is intended to form a community testbed to study the challenges of large scale machine learning and cloud and edge-computing in the context of robotic manipulation. In this work, we describe the mechanical design of the system, its initial software stack and evaluate the repeatability of motions executed by the proposed robot arm design. A local network API throughput and latency analysis is also provided. CloudGripper-Rope-100, a dataset of more than a hundred hours of randomized rope pushing interactions and approximately 4 million camera images is collected and serves as a proof of concept demonstrating data collection capabilities. A project website with more information is available at https://cloudgripper.org.