ACT: Perceptive Acting Under Uncertainty

About the Project

A successful integration of autonomous robots in the real-world depends on them taking socially acceptable actions. The ACT project bridges Neuroscience, Behavioral Psychology, Robotics, and AI to study interactions with humans and autonomous systems and develop new application for safe navigation. Our Lab's role in the project is to create a fundamental understanding of how autonomous agents can cope with uncertainty and demonstrate risk-aware autonomous agents that are demonstrably trustable and predictable.

This project contributed towards improving the safety of autonomous vehicles(AVs) by introducing a method to learn human-like driving behaviours which are then encoded into a local planner. Such behaviors are desired in AVs so that human drivers can trust these systems and also enables smoother coordination with other agents on the road. This behavior is learnt through an interactive imitation learning approach which is data efficient and is then encoded into a model predictive contouring controller. The proposed method is validated using a simulator which demonstrated it's ability to learn human-like driving behaviors.

Another contribution from this project has been to predict human behavior in traffic. This is an important and challenging aspect since often times human behavior is not deterministic. Previous research has often struggled with random uncertainties in human movements which are difficult to model. A novel model called TrajFlow is presented which is capable of fitting distributions present in training samples. A recurrent neural network autoencoder is developed which captures the relevant features of a trajectory and the decoder is able to predict trajectories beyond the length of the training data.

Project Demonstrations

Funding & Partners

This project is funded the the Dutch Research Council NWO-NWA, within the "Acting under Uncertainty (ACT)" project (Grant No. NWA.1292.19.298).

Deploying mobile robots safely among humans requires the motion planner to account for the uncertainty in the other agents' predicted trajectories. This remains challenging in traditional approaches, especially with arbitrarily shaped predictions and real-time constraints. To address these challenges, we propose a Dynamic Risk-Aware Model Predictive Path Integral control (DRA-MPPI), a motion planner that incorporates uncertain future motions modelled with potentially non-Gaussian stochastic predictions. By leveraging MPPI's gradient-free nature, we propose a method that efficiently approximates the joint Collision Probability (CP) among multiple dynamic obstacles for several hundred sampled trajectories in real-time via a Monte Carlo (MC) approach. This enables the rejection of samples exceeding a predefined CP threshold or the integration of CP as a weighted objective within the navigation cost function. Consequently, DRA-MPPI mitigates the freezing robot problem while enhancing safety. Real-world and simulated experiments with multiple dynamic obstacles demonstrate DRA-MPPI's superior performance compared to state-of-the-art approaches, including Scenario-based Model Predictive Control (S-MPC), Frenét planner, and vanilla MPPI.

To safely and efficiently solve motion planning problems in multiagent settings, most approaches attempt to solve a joint optimization that explicitly accounts for the responses triggered in other agents. This often results in solutions with an exponential computational complexity, making these methods intractable for complex scenarios with many agents. While sequential predict-and-plan approaches are more scalable, they tend to perform poorly in highly interactive environments. This paper proposes a method to improve the interactive capabilities of sequential predict-and-plan methods in multi-agent navigation problems by introducing predictability as an optimization objective. We interpret predictability throughnthe use of general prediction models, by allowing agents to predict themselves and estimate how they align with these external predictions. We formally introduce this behavior through the freeenergy of the system, which reduces (under appropriate bounds) to the Kullback-Leibler divergence between plan and prediction, and use this as a penalty for unpredictable trajectories. The proposed interpretation of predictability allows agents to more robustly leverage prediction models, and fosters a ‘soft social convention' that accelerates agreement on coordination strategies without the need of explicit high level control or communication. We show how this predictability-aware planning leads to lower-cost trajectories and reduces planning effort in a set of multi-robot problems, including autonomous driving experiments with human driver data, where we show that the benefits of considering predictability apply even when only the ego-agent uses this strategy.

The estimation of probability density functions is a fundamental problem in science and engineering. However, common methods such as kernel density estimation (KDE) have been demonstrated to lack robustness, while more complex methods have not been evaluated in multi-modal estimation problems. In this paper, we present ROME (RObust Multi-modal Estimator), a non-parametric approach for density estimation which addresses the challenge of estimating multi-modal, non-normal, and highly correlated distributions. ROME utilizes clustering to segment a multi-modal set of samples into multiple uni-modal ones and then combines simple KDE estimates obtained for individual clusters in a single multi-modal estimate. We compared our approach to state-of-the-art methods for density estimation as well as ablations of ROME, showing that it not only outperforms established methods but is also more robust to a variety of distributions. Our results demonstrate that ROME can overcome the issues of over-fitting and over-smoothing exhibited by other estimators.

About the Project

Project Demonstrations

Funding & Partners

Related Publications