Novateur Research Solutions

Novateur Research Solution

Emmanuel Tung

Khurram Hassan-Shafique

Zeeshan Rasheed

Mark Tenzer

Alireza Zaeemzadeh

Eric Chen

An autonomous system for detecting, localizing, and potentially deactivating chemical threats or emissions using multiple sensing modalities and reinforcement learning techniques. The system includes visual sensors (e.g., RGB, RGBD, LIDAR), non-visual sensors (e.g., gas concentration, airflow, GPS, RADAR), a neural network architecture and processor to fuse information from different sensors, a module based on deep reinforcement learning for decision making, and a robotic interface for executing actions. The neural network extracts relevant information from sensor streams and encodes them into a joint embedding space. The module considers the current observations, historical data, and previous actions to determine the optimal action for threat localization under partially observable conditions. The system is trained in simulated environments to minimize source localization time while accounting for various constraints. The autonomous system enables effective chemical threat detection and source localization in complex, dynamic environments without endangering human operators.

Autonomous source localization

Generating trajectories from an implicit neural representation (INR) model to predict human mobility in uncertain traffic conditions includes receiving geocoordinate data representing vehicle motion observations of a traffic pattern; receiving a road network based on the geocoordinate data; training the INR model to learn continuous, latent fields of stochastic traffic properties over space and time based on the geocoordinate data; utilizing the INR model to extract spatio-temporal speed distributions from the geocoordinate data; applying a near-shortest-path, heuristic algorithm, weighted by predictions of the INR model, to produce real-world routing choices for traversing the road network; generating trajectories for transportation between an origin and destination in the road network using the algorithm and the predictions of the INR model, wherein the trajectories reflect non-deterministic and diverse route choices in the road network; and outputting generated trajectories to improve routing choices for a GPS and to provide the route choices for selection.

Generating trajectories from implicit neural models

Growing community of inventors

Ashburn, VA, United States of America

Emmanuel Tung