This paper introduces the Differentiable Algorithm Network (DAN), a composable architecture for robot learning systems. A DAN is composed of neural network modules, each encoding a differentiable robot algorithm and an associated model; and it is trained end-to-end from data. DAN combines the strengths of model-driven modular system design and data-driven end-to-end learning. The algorithms and models act as structural assumptions to reduce the data requirements for learning; end-to-end learning allows the modules to adapt to one another and compensate for imperfect models and algorithms, in order to achieve the best overall system performance. We illustrate the DAN methodology through a case study on a simulated robot system, which learns to navigate in complex 3-D environments with only local visual observations and an image of a partially correct 2-D floor map.

Autonomous driving in a crowded environment, e.g., a busy traffic intersection, is an unsolved challenge for robotics. The robot vehicle must contend with a dynamic and partially observable environment, noisy sensors, and many agents. A principled approach is to formalize it as a Partially Observable Markov Decision Process (POMDP) and solve it through online belief-tree search. To handle a large crowd and achieve real-time performance in this very challenging setting, we propose LeTS-Drive, which integrates online POMDP planning and deep learning. It consists of two phases. In the offline phase, we learn a policy and the corresponding value function by imitating the belief tree search. In the online phase, the learned policy and value function guide the belief tree search. LeTS-Drive leverages the robustness of planning and the runtime efficiency of learning to enhance the performance of both. Experimental results in simulation show that LeTS-Drive outperforms either planning or imitation learning alone and develops sophisticated driving skills.

Scarce data is a major challenge to scaling robot learning to truly complex tasks, as we need to generalize locally learned policies over different task contexts. Contextual policy search offers data-efficient learning and generalization by explicitly conditioning the policy on a parametric context space. In this paper, we further structure the contextual policy representation. We propose to factor contexts into two components: target contexts that describe the task objectives, e.g. target position for throwing a ball; and environment contexts that characterize the environment, e.g. initial position or mass of the ball. Our key observation is that experience can be directly generalized over target contexts. We show that this can be easily exploited in contextual policy search algorithms. In particular, we apply factorization to a Bayesian optimization approach to contextual policy search both in sampling-based and active learning settings. Our simulation results show faster learning and better generalization in various robotic domains. See our supplementary video: https://youtu.be/MNTbBAOufDY.

Trust in autonomy is essential for effective human-robot collaboration and user adoption of autonomous systems such as robot assistants. This paper introduces a computational model which integrates trust into robot decision-making. Specifically, we learn from data a partially observable Markov decision process (POMDP) with human trust as a latent variable. The trust-POMDP model provides a principled approach for the robot to (i) infer the trust of a human teammate through interaction, (ii) reason about the effect of its own actions on human trust, and (iii) choose actions that maximize team performance over the long term. We validated the model through human subject experiments on a table-clearing task in simulation (201 participants) and with a real robot (20 participants). In our studies, the robot builds human trust by manipulating low-risk objects first. Interestingly, the robot sometimes fails intentionally in order to modulate human trust and achieve the best team performance. These results show that the trust-POMDP calibrates trust to improve human-robot team performance over the long term. Further, they highlight that maximizing trust alone does not always lead to the best performance.

We propose to take a novel approach to robot system design where each building block of a larger system is represented as a differentiable program, i.e. a deep neural network. This representation allows for integrating algorithmic planning and deep learning in a principled manner, and thus combine the benefits of model-free and model-based methods. We apply the proposed approach to a challenging partially observable robot navigation task. The robot must navigate to a goal in a previously unseen 3-D environment without knowing its initial location, and instead relying on a 2-D floor map and visual observations from an onboard camera. We introduce the Navigation Networks (NavNets) that encode state estimation, planning and acting in a single, end-to-end trainable recurrent neural network. In preliminary simulation experiments we successfully trained navigation networks to solve the challenging partially observable navigation task.

Particle filters sequentially approximate posterior distributions by sampling representative points and updating them independently. The idea is applied in various domains, e.g. reasoning with uncertainty in robotics. A remaining challenge is constructing probabilistic models of the system, which can be especially hard for complex sensors, e.g. a camera. We introduce the Particle Filter Networks (PF-nets) that encode both a learned probabilistic system model and the particle filter algorithm in a single neural network architecture. The unified representation allows learning models end-to-end, circumventing the difficulties of conventional model-based methods. We applied PF-nets to a challenging visual localization task that requires matching visual features from camera images with the geometry encoded in a 2-D floor map. In preliminary experiments end-to-end PF-nets consistently outperformed alternative learning architectures, as well as conventional model-based methods.

This paper introduces the QMDP-net, a neural network architecture for planning under partial observability. The QMDP-net combines the strengths of model-free learning and model-based planning. It is a recurrent policy network, but it represents a policy for a parameterized set of tasks by connecting a model with a planning algorithm that solves the model, thus embedding the solution structure of planning in a network learning architecture. The QMDP-net is fully differentiable and allows for end-to-end training. We train a QMDP-net on different tasks so that it can generalize to new ones in the parameterized task set and “transfer” to other similar tasks beyond the set. In preliminary experiments, QMDP-net showed strong performance on several robotic tasks in simulation. Interestingly, while QMDP-net encodes the QMDP algorithm, it sometimes outperforms the QMDP algorithm in the experiments, as a result of end-to-end learning.

This paper introduces the QMDP-net, a neural network architecture for planning under partial observability. The QMDP-net combines the strengths of model-free learning and model-based planning. It is a recurrent policy network, but it represents a policy for a parameterized set of tasks by connecting a model with a planning algorithm that solves the model, thus embedding the solution structure of planning in a network learning architecture. The QMDP-net is fully differentiable and allows for end-to-end training. We train a QMDP-net on different tasks so that it can generalize to new ones in the parameterized task set and "transfer" to other similar tasks beyond the set. In preliminary experiments, QMDP-net showed strong performance on several robotic tasks in simulation. Interestingly, while QMDP-net encodes the QMDP algorithm, it sometimes outperforms the QMDP algorithm in the experiments, as a result of end-to-end learning.

The partially observable Markov decision process (POMDP) provides a principled general framework for planning under uncertainty, but solving POMDPs optimally is computationally intractable, due to the "curse of dimensionality" and the "curse of history". To overcome these challenges, we introduce the Determinized Sparse Partially Observable Tree (DESPOT), a sparse approximation of the standard belief tree, for online planning under uncertainty. A DESPOT focuses online planning on a set of randomly sampled scenarios and compactly captures the "execution" of all policies under these scenarios. We show that the best policy obtained from a DESPOT is near-optimal, with a regret bound that depends on the representation size of the optimal policy. Leveraging this result, we give an anytime online planning algorithm, which searches a DESPOT for a policy that optimizes a regularized objective function. Regularization balances the estimated value of a policy under the sampled scenarios and the policy size, thus avoiding overfitting. The algorithm demonstrates strong experimental results, compared with some of the best online POMDP algorithms available. It has also been incorporated into an autonomous driving system for real-time vehicle control. The source code for the algorithm is available online.

Scarce data is a major challenge to scaling robot learning to truly complex tasks, as we need to generalize locally learned policies over different "contexts". Bayesian optimization approaches to contextual policy search (CPS) offer data-efficient policy learning that generalize over a context space. We propose to improve data- efficiency by factoring typically considered contexts into two components: target- type contexts that correspond to a desired outcome of the learned behavior, e.g. target position for throwing a ball; and environment type contexts that correspond to some state of the environment, e.g. initial ball position or wind speed. Our key observation is that experience can be directly generalized over target-type contexts. Based on that we introduce Factored Contextual Policy Search with Bayesian Optimization for both passive and active learning settings. Preliminary results show faster policy generalization on a simulated toy problem.