Agents that can learn to imitate behaviours observed in video -- without having direct access to internal state or action information of the observed agent -- are more suitable for learning in the natural world. However, formulating a reinforcement learning (RL) agent that facilitates this goal remains a significant challenge. We approach this challenge using contrastive training to learn a reward function by comparing an agent's behaviour with a single demonstration. We use a Siamese recurrent neural network architecture to learn rewards in space and time between motion clips while training an RL policy to minimize this distance. Through experimentation, we also find that the inclusion of multi-task data and additional image encoding losses improve the temporal consistency of the learned rewards and, as a result, significantly improve policy learning. We demonstrate our approach on simulated humanoid, dog, and raptor agents in 2D and quadruped and humanoid agents in 3D. We show that our method outperforms current state-of-the-art techniques and can learn to imitate behaviours from a single video demonstration.

Data is the driving force of machine learning, with the amount and quality of training data often being more important for the performance of a system than architecture and training details. But collecting, processing and annotating real data at scale is difficult, expensive, and frequently raises additional privacy, fairness and legal concerns. Synthetic data is a powerful tool with the potential to address these shortcomings: 1) it is cheap 2) supports rich ground-truth annotations 3) offers full control over data and 4) can circumvent or mitigate problems regarding bias, privacy and licensing. Unfortunately, software tools for effective data generation are less mature than those for architecture design and training, which leads to fragmented generation efforts. To address these problems we introduce Kubric, an open-source Python framework that interfaces with PyBullet and Blender to generate photo-realistic scenes, with rich annotations, and seamlessly scales to large jobs distributed over thousands of machines, and generating TBs of data. We demonstrate the effectiveness of Kubric by presenting a series of 13 different generated datasets for tasks ranging from studying 3D NeRF models to optical flow estimation. We release Kubric, the used assets, all of the generation code, as well as the rendered datasets for reuse and modification.

To Perform reliably and consistently over sustained periods of time, large-scale automation critically relies on computer simulation. Simulation allows us and supervisory AI to effectively design, validate, and continuously improve complex processes, and helps practitioners to gain insight into the operation and justify future investments. While numerous successful applications of simulation in industry exist, such as circuit simulation, finite element methods, and computeraided design (CAD), state-of-the-art simulators fall short of accurately modeling physical phenomena, such as friction, impact, and deformation.

Contrastive self-supervised learning has emerged as a promising approach to unsupervised visual representation learning. In general, these methods learn global (image-level) representations that are invariant to different views (i.e., compositions of data augmentation) of the same image. However, many visual understanding tasks require dense (pixel-level) representations. In this paper, we propose View-Agnostic Dense Representation (VADeR) for unsupervised learning of dense representations. VADeR learns pixelwise representations by forcing local features to remain constant over different viewing conditions. Specifically, this is achieved through pixel-level contrastive learning: matching features (that is, features that describes the same location of the scene on different views) should be close in an embedding space, while non-matching features should be apart. VADeR provides a natural representation for dense prediction tasks and transfers well to downstream tasks. Our method outperforms ImageNet supervised pretraining (and strong unsupervised baselines) in multiple dense prediction tasks.

Learning object models in the wild from natural human interactions is an essential ability for robots to perform general tasks. In this paper we present a robocentric multimodal dataset addressing this key challenge. Our dataset focuses on interactions where the user teaches new objects to the robot in various ways. It contains synchronized recordings of visual (3 cameras) and audio data which provide a challenging evaluation framework for different tasks. Additionally, we present an end-to-end system that learns object models using object patches extracted from the recorded natural interactions. Our proposed pipeline follows these steps: (a) recognizing the interaction type, (b) detecting the object that the interaction is focusing on, and (c) learning the models from the extracted data. Our main contribution lies in the steps towards identifying the target object patches of the images. We demonstrate the advantages of combining language and visual features for the interaction recognition and use multiple views to improve the object modelling. Our experimental results show that our dataset is challenging due to occlusions and domain change with respect to typical object learning frameworks. The performance of common out-of-the-box classifiers trained on our data is low. We demonstrate that our algorithm outperforms such baselines.