Applied Intuition’s Automated Parking Development Solution

We are announcing the general availability of Applied Intuition’s automated parking development solution that enables advanced driver-assistance systems (ADAS) and automated driving (AD) engineering teams to develop reliable and efficient automated parking systems (APS). The solution includes simulation and data tools that cover parking operational design domains (ODDs), multi-sensor suites, and test cases, allowing teams to deploy APS up to 12x faster.

The Role of APS in the Automotive Industry

Approximately 20% of all vehicle accidents in the U.S. occur in parking lots. Parking can be stressful, especially in crowded areas like parking structures at malls or in outdoor parking lots at grocery stores. Drivers must simultaneously monitor pedestrians and moving vehicles around the car to ensure their safety while finding and parking in narrow spaces. 

Automated parking systems enable cars to self-park, minimizing the chance of accidents. This increases safety and brings peace of mind to drivers. Despite the benefits, many drivers have historically not relied on their APS because past systems have been unreliable, slow, and limited in their operating range. That is now changing as technologies developed in pursuit of L3 urban and highway ADAS, like multi-sensor fusion and occupancy networks, make it possible to develop APS with improved safety, reliability, range, and speed.

Challenges in Developing APS

OEMs and Tier 1 suppliers face unique challenges when developing APS. First, parking ODDs are extremely diverse, varying across geographical regions, parking infrastructures (e.g. surface lots, underground parking, residential parking), and road markings. Each variation has the potential to bring new edge cases or rare events that challenge the APS and require further development before the system is production-ready.

Additionally, APS need 360-degree sensor and perception coverage with high accuracy at both near and farther-off ranges to ensure that they avoid vulnerable road users (VRUs)—which often are partially occluded by other cars—and park safely in narrow spots. This often requires introducing sensing modalities not used for highway ADAS, like ultrasonic sensors and fisheye cameras, and the fusion of ten or more sensors.

Lastly, planning, prediction, and controls bring their own challenges. Vehicle and pedestrian movement is less structured in parking lots than on regular roads and sidewalks, making it harder to plan or predict motion. Vehicle dynamics also become non-linear at lower speeds and higher steering angles, further complicating the control of vehicle motion in narrow spaces.

These challenges make it impossible to comprehensively develop, test, and validate APS with the same technology and toolchain used for highway ADAS.

Applied Intuition’s Solution

Applied Intuition has developed an automated parking development solution that allows ADAS and AD teams to navigate the aforementioned challenges efficiently and develop reliable classical or machine learning (ML)-based APS. Using this solution brings teams the following benefits:

Enhanced APS safety and reliability

Moving testing earlier in the development process with Sensor Sim, Object Sim and Log Sim—from actual environments to simulated environments—can enhance APS reliability. Teams can test parking edge cases throughout the development process to avoid safety-critical regressions. When regressions are found, ML models can be retrained with synthetic parking data generated to target specific failures, and new implementations can be tested immediately in simulation.

Customized development for APS-specific needs

Capture the nuances of a parking ODD with editable maps available in Map Toolset, spanning different parking infrastructures and regions. Comprehensively test 360-degree sensing and perception with multi-sensor software-in-the-loop (SIL) and hardware-in-the-loop (HIL) simulation that models critical APS sensors like ultrasonics and fisheye cameras. Understand the impact of low speeds and high steering angles on vehicle dynamics with CarSim.

Reduced time to market by developing up to 12x faster

Develop, test, and validate reliable APS up to 12x faster. Our solution offers ready-to-use parking-specific ODD taxonomies, test suites, and maps that enable engineering teams to start simulation-based testing on day one of their development process. Scheduled and pre-commit tests ensure that teams catch and fix regressions before deploying a system on the road. The automated parking development solution also offers programmatically generated synthetic data and ground truth labels (e.g., bounding boxes, depth, semantic segmentation), enabling teams to retrain ML models without data collection and labeling delays.

Contact us today to learn more about Applied Intuition’s automated parking development solution.