Raia Hadsell

Overcoming catastrophic forgetting in neural networks

Significance Deep neural networks are currently the most successful machine-learning technique for solving a variety of tasks, including language translation, image classification, and image generation. One weakness of such models is that, unlike humans, they are unable to learn multiple tasks sequentially. In this work we propose a practical solution to train such models sequentially by protecting the weights important for previous tasks. This approach, inspired by synaptic consolidation in neuroscience, enables state of the art results on multiple reinforcement learning problems experienced sequentially. The ability to learn tasks in a sequential fashion is crucial to the development of artificial intelligence. Until now neural networks have not been capable of this and it has been widely thought that catastrophic forgetting is an inevitable feature of connectionist models. We show that it is possible to overcome this limitation and train networks that can maintain expertise on tasks that they have not experienced for a long time. Our approach remembers old tasks by selectively slowing down learning on the weights important for those tasks. We demonstrate our approach is scalable and effective by solving a set of classification tasks based on a hand-written digit dataset and by learning several Atari 2600 games sequentially.

Deep belief net learning in a long-range vision system for autonomous off-road driving

We present a learning-based approach for long-range vision that is able to accurately classify complex terrain at distances up to the horizon, thus allowing high-level strategic planning. A deep belief network is trained with unsupervised data and a reconstruction criterion to extract features from an input image, and the features are used to train a realtime classifier to predict traversability. The online supervision is given by a stereo module that provides robust labels for nearby areas up to 12 meters distant. The approach was developed and tested on the LAGR mobile robot.

Online Learning for Offroad Robots: Spatial Label Propagation to Learn Long-Range Traversability

We present a solution to the problem of long-range obstacle/path recognition in autonomous robots. The system uses sparse traversability information from a stereo module to train a classifier online. The trained classifier can then predict the traversability of the entire scene. A distance-normalized image pyramid makes it possible to efficiently train on each frame seen by the robot, using large windows that contain contextual information as well as shape, color, and texture. Traversability labels are initially obtained for each target using a stereo module, then propagated to other views of the same target using temporal and spatial concurrences, thus training the classifier to be viewinvariant. A ring buffer simulates short-term memory and ensures that the discriminative learning is balanced and consistent. This long-range obstacle detection system sees obstacles and paths at 30-40 meters, far beyond the maximum stereo range of 12 meters, and adapts very quickly to new environments. Experiments were run on the LAGR robot platform.

Mapping and planning under uncertainty in mobile robots with long-range perception

Recent advances in self-supervised learning have enabled very long-range visual detection of obstacles and pathways (to 100 meters or more). Unfortunately, the category and range of regions at such large distances come with a considerable amount of uncertainty. We present a mapping and planning system that accurately represents range and category uncertainties, and accumulates the evidence from multiple frames in a principled way. The system relies on a hyperbolicpolar map centered on the robot with a 200 m radius. Map cells are histograms that accumulate evidence obtained from a self-supervised object classifier operating on image windows. The performance of the system is demonstrated on the LAGR off-road robot platform.

Vector-based navigation using grid-like representations in artificial agents

Deep neural networks have achieved impressive successes in fields ranging from object recognition to complex games such as Go1,2. Navigation, however, remains a substantial challenge for artificial agents, with deep neural networks trained by reinforcement learning3–5 failing to rival the proficiency of mammalian spatial behaviour, which is underpinned by grid cells in the entorhinal cortex6. Grid cells are thought to provide a multi-scale periodic representation that functions as a metric for coding space7,8 and is critical for integrating self-motion (path integration)6,7,9 and planning direct trajectories to goals (vector-based navigation)7,10,11. Here we set out to leverage the computational functions of grid cells to develop a deep reinforcement learning agent with mammal-like navigational abilities. We first trained a recurrent network to perform path integration, leading to the emergence of representations resembling grid cells, as well as other entorhinal cell types12. We then showed that this representation provided an effective basis for an agent to locate goals in challenging, unfamiliar, and changeable environments—optimizing the primary objective of navigation through deep reinforcement learning. The performance of agents endowed with grid-like representations surpassed that of an expert human and comparison agents, with the metric quantities necessary for vector-based navigation derived from grid-like units within the network. Furthermore, grid-like representations enabled agents to conduct shortcut behaviours reminiscent of those performed by mammals. Our findings show that emergent grid-like representations furnish agents with a Euclidean spatial metric and associated vector operations, providing a foundation for proficient navigation. As such, our results support neuroscientific theories that see grid cells as critical for vector-based navigation7,10,11, demonstrating that the latter can be combined with path-based strategies to support navigation in challenging environments.Grid-like representations emerge spontaneously within a neural network trained to self-localize, enabling the agent to take shortcuts to destinations using vector-based navigation.

Adaptive long range vision in unstructured terrain

A novel probabilistic online learning framework for autonomous off-road robot navigation is proposed. The system is purely vision-based and is particularly designed for predicting traversability in unknown or rapidly changing environments. It uses self-supervised learning to quickly adapt to novel terrains after processing a small number of frames, and it can recognize terrain elements such as paths, man-made structures, and natural obstacles at ranges up to 30 meters. The system is developed on the LAGR mobile robot platform and the performance is evaluated using multiple metrics, including ground truth.

Speed-range dilemmas for vision-based navigation in unstructured terrain

Abstract The performance of vision-based navigation systems for off-road mobile robots depends crucially on the resolution of the camera, the sophistication of the visual processing, the latency between image and sensor capture to actuator control, and the period of the control loop. One particularly important design question is whether one should increase the resolution of the camera images, and the range of the obstacle detection algorithms, at the expense of latency and control loop period. We first report experimental results on the resolution-period trade-off with a stereo vision-based navigation system implemented on the LAGR mobile robot platform. We propose a multi-agent perception and control architecture that combines a sophisticated long-range path detection method operating at high resolution and low frame rate, with a simple stereo-based obstacle detection method operating at low resolution, high frame rate, and low latency. The system combines the advantages of the long-range module for strategic path planning, with the advantages of the short-range module for tactical driving.

The limits and potentials of deep learning for robotics

The application of deep learning in robotics leads to very specific problems and research questions that are typically not addressed by the computer vision and machine learning communities. In this paper we discuss a number of robotics-specific learning, reasoning, and embodiment challenges for deep learning. We explain the need for better evaluation metrics, highlight the importance and unique challenges for deep robotic learning in simulation, and explore the spectrum between purely data-driven and model-driven approaches. We hope this paper provides a motivating overview of important research directions to overcome the current limitations, and helps to fulfill the promising potentials of deep learning in robotics.

Reply to Huszár: The elastic weight consolidation penalty is empirically valid

In our recent work on elastic weight consolidation (EWC) (1) we show that forgetting in neural networks can be alleviated by using a quadratic penalty whose derivation was inspired by Bayesian evidence accumulation. In his letter (2), Dr. Huszar provides an alternative form for this penalty by following the standard work on expectation propagation using the Laplace approximation (3). He correctly argues that in cases when more than two tasks are undertaken the two forms of the penalty are different. Dr. Huszar also shows that for a toy linear regression problem his expression appears to be better. We would like to thank Dr. Huszar for pointing out … [↵][1]1To whom correspondence should be addressed. Email: kirkpatrick@google.com. [1]: #xref-corresp-1-1