CRISP: Object Pose and Shape Estimation with Test-Time Adaptation

We consider the problem of estimating object pose and shape from an RGB-D image. Our first contribution is to introduce CRISP, a category-agnostic object pose and shape estimation pipeline. The pipeline implements an encoder-decoder model for shape estimation. It uses FiLM-conditioning for implicit shape reconstruction and a DPT-based network for estimating pose-normalized points for pose estimation. As a second contribution, we propose an optimization-based pose and shape corrector that can correct estimation errors caused by a domain gap. Observing that the shape decoder is well behaved in the convex hull of known shapes, we approximate the shape decoder with an active shape model, and show that this reduces the shape correction problem to a constrained linear least squares problem, which can be solved efficiently by an interior point algorithm. Third, we introduce a self-training pipeline to perform self-supervised domain adaptation of CRISP. The self-training is based on a correct-and-certify approach, which leverages the corrector to generate pseudo-labels at test time, and uses them to self-train CRISP. We demonstrate CRISP (and the self-training) on YCBV, SPE3R, and NOCS datasets. CRISP shows high performance on all the datasets. Moreover, our self-training is capable of bridging a large domain gap. Finally, CRISP also shows an ability to generalize to unseen objects. Code and pre-trained models will be available on https://web.mit.edu/sparklab/research/crisp_object_pose_shape/.

Test-Time Certifiable Self-Supervision to Bridge the Sim2Real Gap in Event-Based Satellite Pose Estimation

Deep learning plays a critical role in vision-based satellite pose estimation. However, the scarcity of real data from the space environment means that deep models need to be trained using synthetic data, which raises the Sim2Real domain gap problem. A major cause of the Sim2Real gap are novel lighting conditions encountered during test time. Event sensors have been shown to provide some robustness against lighting variations in vision-based pose estimation. However, challenging lighting conditions due to strong directional light can still cause undesirable effects in the output of commercial off-the-shelf event sensors, such as noisy/spurious events and inhomogeneous event densities on the object. Such effects are non-trivial to simulate in software, thus leading to Sim2Real gap in the event domain. To close the Sim2Real gap in event-based satellite pose estimation, the paper proposes a test-time self-supervision scheme with a certifier module. Self-supervision is enabled by an optimisation routine that aligns a dense point cloud of the predicted satellite pose with the event data to attempt to rectify the inaccurately estimated pose. The certifier attempts to verify the corrected pose, and only certified test-time inputs are backpropagated via implicit differentiation to refine the predicted landmarks, thus improving the pose estimates and closing the Sim2Real gap. Results show that the our method outperforms established test-time adaptation schemes.

Safe Distributed Control of Multi-Robot Systems with Communication Delays

Safe operation of multi-robot systems is critical, especially in communication-degraded environments such as underwater for seabed mapping, underground caves for navigation, and in extraterrestrial missions for assembly and construction. We address safety of networked autonomous systems where the information exchanged between robots incurs communication delays. We formalize a notion of distributed control barrier function (CBF) for multi-robot systems, a safety certificate amenable to a distributed implementation, which provides formal ground to using graph neural networks to learn safe distributed controllers. Further, we observe that learning a distributed controller ignoring delays can severely degrade safety. Our main contribution is a predictor-based framework to train a safe distributed controller under communication delays, where the current state of nearby robots is predicted from received data and age-of-information. Numerical experiments on multi-robot collision avoidance show that our predictor-based approach can significantly improve the safety of a learned distributed controller under communication delays

Foundations of Spatial Perception for Robotics: Hierarchical Representations and Real-time Systems

3D spatial perception is the problem of building and maintaining an actionable and persistent representation of the environment in real-time using sensor data and prior knowledge. Despite the fast-paced progress in robot perception, most existing methods either build purely geometric maps (as in traditional SLAM) or flat metric-semantic maps that do not scale to large environments or large dictionaries of semantic labels. The first part of this paper is concerned with representations: we show that scalable representations for spatial perception need to be hierarchical in nature. Hierarchical representations are efficient to store, and lead to layered graphs with small treewidth, which enable provably efficient inference. We then introduce an example of hierarchical representation for indoor environments, namely a 3D scene graph, and discuss its structure and properties. The second part of the paper focuses on algorithms to incrementally construct a 3D scene graph as the robot explores the environment. Our algorithms combine 3D geometry, topology (to cluster the places into rooms), and geometric deep learning (e.g., to classify the type of rooms the robot is moving across). The third part of the paper focuses on algorithms to maintain and correct 3D scene graphs during long-term operation. We propose hierarchical descriptors for loop closure detection and describe how to correct a scene graph in response to loop closures, by solving a 3D scene graph optimization problem. We conclude the paper by combining the proposed perception algorithms into Hydra, a real-time spatial perception system that builds a 3D scene graph from visual-inertial data in real-time. We showcase Hydra's performance in photo-realistic simulations and real data collected by a Clearpath Jackal robots and a Unitree A1 robot. We release an open-source implementation of Hydra at https://github.com/MIT-SPARK/Hydra.

A Correct-and-Certify Approach to Self-Supervise Object Pose Estimators via Ensemble Self-Training

Real-world robotics applications demand object pose estimation methods that work reliably across a variety of scenarios. Modern learning-based approaches require large labeled datasets and tend to perform poorly outside the training domain. Our first contribution is to develop a robust corrector module that corrects pose estimates using depth information, thus enabling existing methods to better generalize to new test domains; the corrector operates on semantic keypoints (but is also applicable to other pose estimators) and is fully differentiable. Our second contribution is an ensemble self-training approach that simultaneously trains multiple pose estimators in a self-supervised manner. Our ensemble self-training architecture uses the robust corrector to refine the output of each pose estimator; then, it evaluates the quality of the outputs using observable correctness certificates; finally, it uses the observably correct outputs for further training, without requiring external supervision. As an additional contribution, we propose small improvements to a regression-based keypoint detection architecture, to enhance its robustness to outliers; these improvements include a robust pooling scheme and a robust centroid computation. Experiments on the YCBV and TLESS datasets show the proposed ensemble self-training outperforms fully supervised baselines while not requiring 3D annotations on real data.

PyPose: A Library for Robot Learning with Physics-based Optimization

Deep learning has had remarkable success in robotic perception, but its data-centric nature suffers when it comes to generalizing to ever-changing environments. By contrast, physics-based optimization generalizes better, but it does not perform as well in complicated tasks due to the lack of high-level semantic information and the reliance on manual parametric tuning. To take advantage of these two complementary worlds, we present PyPose: a robotics-oriented, PyTorch-based library that combines deep perceptual models with physics-based optimization techniques. Our design goal for PyPose is to make it user-friendly, efficient, and interpretable with a tidy and well-organized architecture. Using an imperative style interface, it can be easily integrated into real-world robotic applications. Besides, it supports parallel computing of any order gradients of Lie groups and Lie algebras and $2^{\text{nd}}$-order optimizers, such as trust region methods. Experiments show that PyPose achieves 3-20$\times$ speedup in computation compared to state-of-the-art libraries. To boost future research, we provide concrete examples across several fields of robotics, including SLAM, inertial navigation, planning, and control.

Leveraging Large Language Models for Robot 3D Scene Understanding

Semantic 3D scene understanding is a problem of critical importance in robotics. While significant advances have been made in spatial perception, robots are still far from having the common-sense knowledge about household objects and locations of an average human. We thus investigate the use of large language models to impart common sense for scene understanding. Specifically, we introduce three paradigms for leveraging language for classifying rooms in indoor environments based on their contained objects: (i) a zero-shot approach, (ii) a feed-forward classifier approach, and (iii) a contrastive classifier approach. These methods operate on 3D scene graphs produced by modern spatial perception systems. We then analyze each approach, demonstrating notable zero-shot generalization and transfer capabilities stemming from their use of language. Finally, we show these approaches also apply to inferring building labels from contained rooms and demonstrate our zero-shot approach on a real environment. All code can be found at https://github.com/MIT-SPARK/llm_scene_understanding.

Correct and Certify: A New Approach to Self-Supervised 3D-Object Perception

We consider an object pose estimation and model fitting problem, where - given a partial point cloud of an object - the goal is to estimate the object pose by fitting a CAD model to the sensor data. We solve this problem by combining (i) a semantic keypoint-based pose estimation model, (ii) a novel self-supervised training approach, and (iii) a certification procedure, that not only verifies whether the output produced by the model is correct or not, but also flags uniqueness of the produced solution. The semantic keypoint detector model is initially trained in simulation and does not perform well on real-data due to the domain gap. Our self-supervised training procedure uses a corrector and a certification module to improve the detector. The corrector module corrects the detected keypoints to compensate for the domain gap, and is implemented as a declarative layer, for which we develop a simple differentiation rule. The certification module declares whether the corrected output produced by the model is certifiable (i.e. correct) or not. At each iteration, the approach optimizes over the loss induced only by the certifiable input-output pairs. As training progresses, we see that the fraction of outputs that are certifiable increases, eventually reaching near $100\%$ in many cases. We also introduce the notion of strong certifiability wherein the model can determine if the predicted object model fit is unique or not. The detected semantic keypoints help us implement this in the forward pass. We conduct extensive experiments to evaluate the performance of the corrector, the certification, and the proposed self-supervised training using the ShapeNet and YCB datasets, and show the proposed approach achieves performance comparable to fully supervised baselines while not requiring pose or keypoint supervision on real data.

Extracting Zero-shot Common Sense from Large Language Models for Robot 3D Scene Understanding

Semantic 3D scene understanding is a problem of critical importance in robotics. While significant advances have been made in simultaneous localization and mapping algorithms, robots are still far from having the common sense knowledge about household objects and their locations of an average human. We introduce a novel method for leveraging common sense embedded within large language models for labelling rooms given the objects contained within. This algorithm has the added benefits of (i) requiring no task-specific pre-training (operating entirely in the zero-shot regime) and (ii) generalizing to arbitrary room and object labels, including previously-unseen ones -- both of which are highly desirable traits in robotic scene understanding algorithms. The proposed algorithm operates on 3D scene graphs produced by modern spatial perception systems, and we hope it will pave the way to more generalizable and scalable high-level 3D scene understanding for robotics.

Visual Navigation for Autonomous Vehicles: An Open-source Hands-on Robotics Course at MIT

This paper reports on the development, execution, and open-sourcing of a new robotics course at MIT. The course is a modern take on "Visual Navigation for Autonomous Vehicles" (VNAV) and targets first-year graduate students and senior undergraduates with prior exposure to robotics. VNAV has the goal of preparing the students to perform research in robotics and vision-based navigation, with emphasis on drones and self-driving cars. The course spans the entire autonomous navigation pipeline; as such, it covers a broad set of topics, including geometric control and trajectory optimization, 2D and 3D computer vision, visual and visual-inertial odometry, place recognition, simultaneous localization and mapping, and geometric deep learning for perception. VNAV has three key features. First, it bridges traditional computer vision and robotics courses by exposing the challenges that are specific to embodied intelligence, e.g., limited computation and need for just-in-time and robust perception to close the loop over control and decision making. Second, it strikes a balance between depth and breadth by combining rigorous technical notes (including topics that are less explored in typical robotics courses, e.g., on-manifold optimization) with slides and videos showcasing the latest research results. Third, it provides a compelling approach to hands-on robotics education by leveraging a physical drone platform (mostly suitable for small residential courses) and a photo-realistic Unity-based simulator (open-source and scalable to large online courses). VNAV has been offered at MIT in the Falls of 2018-2021 and is now publicly available on MIT OpenCourseWare (OCW).