Next-generation combat vehicle software developed for ground maneuver

Approach

Applied Intuition Defense brings commercial best practices and an off-road autonomy stack to leading ground vehicle autonomy programs in the U.S. Army, including the Robotic Combat Vehicle (RCV), the XM30 Mechanized Infantry Combat Vehicle (MICV), and more.

Perception

Build systems designed to understand terrain and surroundings

Use sensor simulation, real-world data, and high-realism synthetic data to train computer vision algorithms. Quickly identify and classify terrain, objects, and friendly or adversary systems near the vehicle. Easily vary environmental parameters like weather, time of day, and lighting for comprehensive testing.

Build and test systems that quickly and accurately understand their surroundings.

Visualize real-world drive data

Ingest, visualize, and query time series sensor data from real-world drives to identify stack performance issues and root causes, and prioritize areas for improvement. Build curated datasets for model retraining.

Generate synthetic data

Generate high-realism, physically accurate synthetic data with pixel-level annotations to augment real-world drive logs and expand training data.

Train models on real and synthetic data

Combine real-world and synthetic data to increase coverage and ensure that perception models are able to accurately and reliably identify and classify object classes across domains, including new and rare threat systems.

Maneuver

Safely model complex real-world interactions

Accelerate planning and controls module development with scenario-based testing. Utilize object-level simulations to test and validate planning and controls module performance in relevant operational scenarios, and trace each test scenario back to program requirements.

Accelerate development by testing common and rare scenarios in physically accurate virtual environments.

Establish your operational design domain (ODD)

Establish initial requirements for system performance and safety, including vehicle maneuvers, terrain types, environmental conditions, and object and obstacle classes.

Model your domain

Create your test environment. Ingest HD maps and generate physically accurate 3D worlds. Add and vary actors, obstacles, and model terrain. Import physics-based vehicle dynamics models.

Run scenario-based tests

Create complex scenarios that respond to program requirements, operational needs, and model interactions with other vehicles and terrain. Vary weather, time, lighting, terrain, and other parameters to comprehensively test system performance. Run large sets of tests at scale in cloud simulation.

T&E

Evaluate the performance of multiple stacks

Compare the performance of multiple autonomy stacks against the same ODD or mission set. Evaluate the performance of one or multiple stacks across common and rare scenarios, and quantify perception, planning, and controls sub-system performance to inform program decisions.

Make program decisions based on quantitative measures of performance.

Define program requirements

Establish initial requirements for system performance and safety, including vehicle maneuvers, terrain types, environmental conditions, and object and obstacle classes.

Define test plans

Define scenarios and test plans to test specific features, integrations, or system-level functionalities for one or multiple stacks to ensure sufficient coverage of program requirements.

Execute test plans

Identify relevant test cases from large parameter spaces. Create and execute scenarios in simulation to save time and costs compared to real-world testing.

Accelerate ground vehicle autonomy

Learn how Applied Intuition Defense can address your autonomy challenges.

Contact us