Learning world models can teach an agent how the world works in an unsupervised manner. Even though it can be viewed as a special case of sequence modeling, progress for scaling world models on robotic applications such as autonomous driving has been somewhat less rapid than scaling language models with Generative Pre-trained Transformers (GPT). We identify two reasons as major bottlenecks: dealing with complex and unstructured observation space, and having a scalable generative model. Consequently, we propose Copilot4D, a novel world modeling approach that first tokenizes sensor observations with VQVAE, then predicts the future via discrete diffusion. To efficiently decode and denoise tokens in parallel, we recast Masked Generative Image Transformer as discrete diffusion and enhance it with a few simple changes, resulting in notable improvement. When applied to learning world models on point cloud observations, Copilot4D reduces prior SOTA Chamfer distance by more than 65% for 1s prediction, and more than 50% for 3s prediction, across NuScenes, KITTI Odometry, and Argoverse2 datasets. Our results demonstrate that discrete diffusion on tokenized agent experience can unlock the power of GPT-like unsupervised learning for robotics.

Reconstructing objects from real world data and rendering them at novel views is critical to bringing realism, diversity and scale to simulation for robotics training and testing. In this work, we present NeuSim, a novel approach that estimates accurate geometry and realistic appearance from sparse in-the-wild data captured at distance and at limited viewpoints. Towards this goal, we represent the object surface as a neural signed distance function and leverage both LiDAR and camera sensor data to reconstruct smooth and accurate geometry and normals. We model the object appearance with a robust physics-inspired reflectance representation effective for in-the-wild data. Our experiments show that NeuSim has strong view synthesis performance on challenging scenarios with sparse training views. Furthermore, we showcase composing NeuSim assets into a virtual world and generating realistic multi-sensor data for evaluating self-driving perception models. The supplementary material can be found at the project website: https://waabi.ai/research/neusim/

3D object detection is a key component of many robotic applications such as self-driving vehicles. While many approaches rely on expensive 3D sensors such as LiDAR to produce accurate 3D estimates, methods that exploit stereo cameras have recently shown promising results at a lower cost. Existing approaches tackle this problem in two steps: first depth estimation from stereo images is performed to produce a pseudo LiDAR point cloud, which is then used as input to a 3D object detector. However, this approach is suboptimal due to the representation mismatch, as the two tasks are optimized in two different metric spaces. In this paper we propose a model that unifies these two tasks and performs them in the same metric space. Specifically, we directly construct a pseudo LiDAR feature volume (PLUME) in 3D space, which is then used to solve both depth estimation and object detection tasks. Our approach achieves state-of-the-art performance with much faster inference times when compared to existing methods on the challenging KITTI benchmark [1].

In this paper, we propose PolyTransform, a novel instance segmentation algorithm that produces precise, geometry-preserving masks by combining the strengths of prevailing segmentation approaches and modern polygon-based methods. In particular, we first exploit a segmentation network to generate instance masks. We then convert the masks into a set of polygons that are then fed to a deforming network that transforms the polygons such that they better fit the object boundaries. Our experiments on the challenging Cityscapes dataset show that our PolyTransform significantly improves the performance of the backbone instance segmentation network and ranks 1st on the Cityscapes test-set leaderboard. We also show impressive gains in the interactive annotation setting.

We tackle the problem of joint perception and motion forecasting in the context of self-driving vehicles. Towards this goal we propose PnPNet, an end-to-end model that takes as input sequential sensor data, and outputs at each time step object tracks and their future trajectories. The key component is a novel tracking module that generates object tracks online from detections and exploits trajectory level features for motion forecasting. Specifically, the object tracks get updated at each time step by solving both the data association problem and the trajectory estimation problem. Importantly, the whole model is end-to-end trainable and benefits from joint optimization of all tasks. We validate PnPNet on two large-scale driving datasets, and show significant improvements over the state-of-the-art with better occlusion recovery and more accurate future prediction.

In this paper we propose to exploit multiple related tasks for accurate multi-sensor 3D object detection. Towards this goal we present an end-to-end learnable architecture that reasons about 2D and 3D object detection as well as ground estimation and depth completion. Our experiments show that all these tasks are complementary and help the network learn better representations by fusing information at various levels. Importantly, our approach leads the KITTI benchmark on 2D, 3D and bird's eye view object detection, while being real-time.

In this paper we tackle the problem of scene flow estimation in the context of self-driving. We leverage deep learning techniques as well as strong priors as in our application domain the motion of the scene can be composed by the motion of the robot and the 3D motion of the actors in the scene. We formulate the problem as energy minimization in a deep structured model, which can be solved efficiently in the GPU by unrolling a Gaussian-Newton solver. Our experiments in the challenging KITTI scene flow dataset show that we outperform the state-of-the-art by a very large margin, while being 800 times faster.

In this paper, we propose a unified panoptic segmentation network (UPSNet) for tackling the newly proposed panoptic segmentation task. On top of a single backbone residual network, we first design a deformable convolution based semantic segmentation head and a Mask R-CNN style instance segmentation head which solve these two subtasks simultaneously. More importantly, we introduce a parameter-free panoptic head which solves the panoptic segmentation via pixel-wise classification. It first leverages the logits from the previous two heads and then innovatively expands the representation for enabling prediction of an extra unknown class which helps better resolving the conflicts between semantic and instance segmentation. Besides, it handles the challenge caused by the varying number of instances and permits back propagation to the bottom modules in an end-to-end manner. Extensive experimental results on Cityscapes, COCO and our internal dataset demonstrate that our UPSNet achieves state-of-the-art performance with much faster inference. Code has been made available at: https://github.com/uber-research/UPSNet.

Our goal is to significantly speed up the runtime of current state-of-the-art stereo algorithms to enable real-time inference. Towards this goal, we developed a differentiable PatchMatch module that allows us to discard most disparities without requiring full cost volume evaluation. We then exploit this representation to learn which range to prune for each pixel. By progressively reducing the search space and effectively propagating such information, we are able to efficiently compute the cost volume for high likelihood hypotheses and achieve savings in both memory and computation.Finally, an image guided refinement module is exploited to further improve the performance. Since all our components are differentiable, the full network can be trained end-to-end. Our experiments show that our method achieves competitive results on KITTI and SceneFlow datasets while running in real-time at 62ms.