Generative models are emerging as promising tools in robotics and reinforcement learning. Yet, even though tasks in these domains typically involve distinct objects, most state-of-the-art methods do not explicitly capture the compositional nature of visual scenes. Two exceptions, MONet and IODINE, decompose scenes into objects in an unsupervised fashion via a set of latent variables. Their underlying generative processes, however, do not account for component interactions. Hence, neither of them allows for principled sampling of coherent scenes. Here we present GENESIS, the first object-centric generative model of visual scenes capable of both decomposing and generating complete scenes by explicitly capturing relationships between scene components. GENESIS parameterises a spatial GMM over pixels which is encoded by component-wise latent variables that are inferred sequentially or sampled from an autoregressive prior. We train GENESIS on two publicly available datasets and probe the information in the latent representations through a set of classification tasks, outperforming several baselines.
The majority of contemporary object-tracking approaches used in autonomous vehicles do not model interactions between objects. This contrasts with the fact that objects' paths are not independent: a cyclist might abruptly deviate from a previously planned trajectory in order to avoid colliding with a car. Building upon HART, a neural, class-agnostic single-object tracker, we introduce a multi-object tracking method MOHART capable of relational reasoning. Importantly, the entire system, including the understanding of interactions and relations between objects, is class-agnostic and learned simultaneously in an end-to-end fashion. We find that the addition of relational-reasoning capabilities to HART leads to consistent performance gains in tracking as well as future trajectory prediction on several real-world datasets (MOTChallenge, UA-DETRAC, and Stanford Drone dataset), particularly in the presence of ego-motion, occlusions, crowded scenes, and faulty sensor inputs. Finally, based on controlled simulations, we propose that a comparison of MOHART and HART may be used as a novel way to measure the degree to which the objects in a video depend upon each other as they move together through time.
Recent work on the representation of functions on sets has considered the use of summation in a latent space to enforce permutation invariance. In particular, it has been conjectured that the dimension of this latent space may remain fixed as the cardinality of the sets under consideration increases. However, we demonstrate that the analysis leading to this conjecture requires mappings which are highly discontinuous and argue that this is only of limited practical use. Motivated by this observation, we prove that an implementation of this model via continuous mappings (as provided by e.g. neural networks or Gaussian processes) actually imposes a constraint on the dimensionality of the latent space. Practical universal function representation for set inputs can only be achieved with a latent dimension at least the size of the maximum number of input elements.
Radar presents a promising alternative to lidar and vision in autonomous vehicle applications, being able to detect objects at long range under a variety of weather conditions. However, distinguishing between occupied and free space from a raw radar scan is notoriously difficult. We consider the challenge of learning an Inverse Sensor Model (ISM) mapping a raw radar observation to occupancy probabilities in a discretised space. We frame this problem as a segmentation task, utilising a deep neural network that is able to learn an inherently probabilistic ISM from raw sensor data considers scene context. In doing so our approach explicitly accounts for the heteroscedastic aleatoric uncertainty for radar that arises due to complex interactions between occlusion and sensor noise. Our network is trained using only partial occupancy labels generated from lidar and able to successfully distinguish between occupied and free space. We evaluate our approach on five hours of data recorded in a dynamic urban environment and show that it significantly outperforms classical constant false-alarm rate (CFAR) filtering approaches in light of challenging noise artefacts whilst identifying space that is inherently uncertain because of occlusion.
We propose an efficient way to output better calibrated uncertainty scores from neural networks. The Distilled Dropout Network (DDN) makes standard (non-Bayesian) neural networks more introspective by adding a new training loss which prevents them from being overconfident. Our method is more efficient than Bayesian neural networks or model ensembles which, despite providing more reliable uncertainty scores, are more cumbersome to train and slower to test. We evaluate DDN on the the task of image classification on the CIFAR-10 dataset and show that our calibration results are competitive even when compared to 100 Monte Carlo samples from a dropout network while they also increase the classification accuracy. We also propose better calibration within the state of the art Faster R-CNN object detection framework and show, using the COCO dataset, that DDN helps train better calibrated object detectors.
Many advanced Learning from Demonstration (LfD) methods consider the decomposition of complex, real-world tasks into simpler sub-tasks. By reusing the corresponding sub-policies within and between tasks, they provide training data for each policy from different high-level tasks and compose them to perform novel ones. Existing approaches to modular LfD focus either on learning a single high-level task or depend on domain knowledge and temporal segmentation. In contrast, we propose a weakly supervised, domain-agnostic approach based on task sketches, which include only the sequence of sub-tasks performed in each demonstration. Our approach simultaneously aligns the sketches with the observed demonstrations and learns the required sub-policies. This improves generalisation in comparison to separate optimisation procedures. We evaluate the approach on multiple domains, including a simulated 3D robot arm control task using purely image-based observations. The results show that our approach performs commensurately with fully supervised approaches, while requiring significantly less annotation effort.
Physical intuition is pivotal for intelligent agents to perform complex tasks. In this paper we investigate the passive acquisition of an intuitive understanding of physical principles as well as the active utilisation of this intuition in the context of generalised object stacking. To this end, we provide: a simulation-based dataset featuring 20,000 stack configurations composed of a variety of elementary geometric primitives richly annotated regarding semantics and structural stability. We train visual classifiers for binary stability prediction on the ShapeStacks data and scrutinise their learned physical intuition. Due to the richness of the training data our approach also generalises favourably to real-world scenarios achieving state-of-the-art stability prediction on a publicly available benchmark of block towers. We then leverage the physical intuition learned by our model to actively construct stable stacks and observe the emergence of an intuitive notion of stackability - an inherent object affordance - induced by the active stacking task. Our approach performs well even in challenging conditions where it considerably exceeds the stack height observed during training or in cases where initially unstable structures must be stabilised via counterbalancing.
Model interpretability and systematic, targeted model adaptation present central tenets in machine learning for addressing limited or biased datasets. In this paper, we introduce neural stethoscopes as a framework for quantifying the degree of importance of specific factors of influence in deep networks as well as for actively promoting and suppressing information as appropriate. In doing so we unify concepts from multitask learning as well as training with auxiliary and adversarial losses. We showcase the efficacy of neural stethoscopes in an intuitive physics domain. Specifically, we investigate the challenge of visually predicting stability of block towers and demonstrate that the network uses visual cues which makes it susceptible to biases in the dataset. Through the use of stethoscopes we interrogate the accessibility of specific information throughout the network stack and show that we are able to actively de-bias network predictions as well as enhance performance via suitable auxiliary and adversarial stethoscope losses.
We present Sequential Attend, Infer, Repeat (SQAIR), an interpretable deep generative model for videos of moving objects. It can reliably discover and track objects throughout the sequence of frames, and can also generate future frames conditioning on the current frame, thereby simulating expected motion of objects. This is achieved by explicitly encoding object presence, locations and appearances in the latent variables of the model. SQAIR retains all strengths of its predecessor, Attend, Infer, Repeat (AIR, Eslami et. al., 2016), including learning in an unsupervised manner, and addresses its shortcomings. We use a moving multi-MNIST dataset to show limitations of AIR in detecting overlapping or partially occluded objects, and show how SQAIR overcomes them by leveraging temporal consistency of objects. Finally, we also apply SQAIR to real-world pedestrian CCTV data, where it learns to reliably detect, track and generate walking pedestrians with no supervision.
We present a self-supervised approach to ignoring "distractors" in camera images for the purposes of robustly estimating vehicle motion in cluttered urban environments. We leverage offline multi-session mapping approaches to automatically generate a per-pixel ephemerality mask and depth map for each input image, which we use to train a deep convolutional network. At run-time we use the predicted ephemerality and depth as an input to a monocular visual odometry (VO) pipeline, using either sparse features or dense photometric matching. Our approach yields metric-scale VO using only a single camera and can recover the correct egomotion even when 90% of the image is obscured by dynamic, independently moving objects. We evaluate our robust VO methods on more than 400km of driving from the Oxford RobotCar Dataset and demonstrate reduced odometry drift and significantly improved egomotion estimation in the presence of large moving vehicles in urban traffic.