Driver Assistance Systems Parking Made Easy—Also Thanks to Optimized Chips

Related Vendors

Renesas Electronics Europe GmbH

FAULHABER Drive Systems

Wuxi InfiMotion Technology Co., Ltd.

Many drivers get stressed again and again when they have to park their vehicle. They do it because the parking space is very narrow or because the traffic behind them is building up pressure. Automated parking assistance systems are intended to provide a remedy.

For many drivers, parking is a nerve-wracking affair—whether it's squeezing into a parking space at the supermarket or maneuvering through narrow parking garages in the city center. There is even a term for the fear of parallel parking: parallelophobia. The Chinese company iMotion wants to reduce the stress of parking. With the help of a highly automated function, vehicles should be able to park themselves in less than 30 seconds—even in crowded or poorly marked areas. Walt Song, who founded iMotion Automotive Technology in 2016, has an easy-to-understand philosophy: "Make everything easy". This applies not only to the driver, but also to the car manufacturer. The goal? To bring functions that were originally developed for luxury vehicles to more affordable mid-range and entry-level models.

Parking Automation Also for Affordable Entry-Level Models

Automated Parking Assist (APA) systems not only promise less stress for the driver, but also fewer fender benders and kerb bumps. APA also allows more vehicles to be parked in a smaller space. However, for car manufacturers focusing on entry-level models, APA—like many other ADAS functions—is proving to be technically complex and cost-intensive. iMotion's integrated driving technology aims to counteract these hurdles.

The APA system guides drivers up to three kilometers along previously learned routes to their destination. With the help of cameras and ultrasonic sensors, it recognizes the surroundings in real time and guides the vehicle into the desired parking space. The technology works in vehicles with electric and combustion engines as well as in various vehicle types—including cars, trucks and off-road vehicles.

What sets iMotion apart from its competitors is its fully integrated system design. While many ADAS manufacturers focus solely on software or hardware, iMotion offers a vertically integrated stack—from algorithm development to middleware and embedded hardware. This is intended to optimize design resources, performance and costs.

At its heart is the intelligent, integrated iMotion iDC500 domain controller. It comprises an APA system based on 360-degree environment recognition and a TCP feature (Trained Commuting Pilot) with self-learning mapping technology. This navigates by "memorizing" certain routes and locally recorded environmental data.

360-degree environment recognition is made possible by vehicle cameras and iMotion's Bird's Eye View technology. This technology generates a bird's eye view of the surroundings to obtain a 3D map of the outside world. The mapping is performed by an AI-based neural network architecture, a so-called transformer model. This combines different camera images to generate a 3D representation of the environment. Competing models in this category are said to have significantly larger parameter scales and require more expensive AI processors with at least 100 TOPS of computing power. The developers at iMotion were able to optimize their algorithms so that they run on the more economical R-Car V4H chipset from Renesas with 34 TOPS—with the same performance. This efficiency gain makes it possible to offer functions such as automated parking and high-resolution, map-based highway piloting. This allows them to align their vehicles with both the expectations of end users and the regulatory safety requirements in Asia and Europe.

Cost-Efficient Parking Platform

In order to successfully integrate high-end ADAS functions into cost-effective hardware, the iMotion team had to overcome various challenges. Advanced sensor, planning and control functions had to be integrated into a compact computer platform. Strict performance and security requirements had to be taken into account.

One of the hurdles was to implement BEV transformers and Occupancy Grid Models (OCC) in Renesas' R-Car V4H architecture. BEV and OCC are a form of spatial representation often used in mobile robotics. iMotion uses a combination of different technologies:

• in "Lightweight" decomposition AI framework breaks down complex Transformer models into smaller, more manageable components to facilitate processing.
• Zero-copy memory scheduling reduces redundant data copying processes when transferring between different memory areas and thus reduces the CPU load.
• Hardware-optimized operator rewriting enables high-precision, centimeter-accurate grid resolution.

For this purpose, computing operations are converted into more efficient formats that are better suited for processing by GPUs, CPUs and specialized AI accelerators.

Another challenge was to ensure system stability and functional safety across different markets and vehicle platforms—from cars and motorhomes to trucks. The modular, scalable design of the iMotion platform ensures global compatibility and robustness through extensive simulations and testing.

Partnership With Renesas: Practice-Oriented Design

The strategic collaboration between iMotion and Renesas has been in place since 2020. The R-Car V4H in particular would have impressed with its balance of computing power, energy efficiency and scalability. The R-Car V4H system-on-chip (SoC) is manufactured using the 7 nm CMOS process and contains an integrated Arm Cortex-R52 core, which makes an external microcontroller obsolete. This increases the efficiency of the chip and reduces power consumption, PCB size and BOM costs. The SoC also integrates a CNN-IP AI accelerator, multiple DSPs and image processing engines optimized for iMotion's software stack.

The device is designed for ADAS and automated driving functions and delivers up to 34 TOPS of deep learning performance. It enables ultra-fast image recognition from camera, radar and lidar data, supporting functions such as surround view perception and realistic 3D visualization for autonomous parking. Highly integrated for cost-efficient electronic control units (ECUs), the R-Car V4H meets NCAP 2025 standards and supports Level 2+ autonomy. A dual-chip configuration ensures uninterrupted upgrades and fail-safe operation, which is essential for Level 3 systems.

It is not only the hardware that plays an important role in the partnership, but also the close cooperation with the Renesas engineering team. "We have worked hand in hand with Renesas for two years," explains company founder Walt Song. "The application teams in China and elsewhere have helped us solve critical bandwidth and integration challenges. This high level of collaboration was critical to bringing the iDC500 to market."

New Objectives

iMotion is continuously developing its technology portfolio through the commercialization of additional, complementary technologies such as adaptive cruise control, lane departure warning and collision warning systems as well as driver fatigue detection. The company plans to integrate its new driving assistance algorithm into series production vehicles this year. This is an important step on the way to complete end-to-end navigation from highway assistance to urban door-to-door guidance.

iMotion expands in Europe and begins market launch in Japan and the ASEAN region. The company is leveraging Renesas' global reach and local service networks to make vehicles safer, smarter and more intuitive. Song explains that this will enable a world in which autonomous vehicles become a mobile extension of the home and workplace: "In the future, autonomous mobility will transform the vehicle interior into a second living space. In the future, it will no longer just be about driving; we will also be able to read, relax or network. Our goal is to make this future a reality—safely, affordably and quickly." (se)

Pu Zhang, Sr Principal Engineer, High Performance Computing, Renesas

Secure and Compliant Authentication in Laboratories

Secure & Contactless User Authentication at GMP/GLP