Automated Valet Parking (AVP)

Automated Valet Parking (AVP) refers to a Level 4 (L4) low-speed autonomous driving technology where the driver simply alights at a designated drop-off point, leaving the vehicle to autonomously enter a car park, locate a vacant parking bay, and park itself without anyone on board. For vehicle retrieval, it can likewise be summoned remotely to a designated pick-up point. Representing the pinnacle of automation in intelligent parking, this technology ultimately aims to replace conventional manual valet parking services.

Working Principle

A typical AVP operational workflow comprises four key stages: Vehicle connectivity (the user sends remote commands via a smartphone app), Bay search (the vehicle autonomously scouts the car park for a vacant spot), Autonomous parking (the vehicle plots its path and reverses into the bay), and Remote summon (the vehicle departs the parking bay and navigates to the designated pick-up point).

From an architectural standpoint, the AVP system operates on a closed-loop "perception-decision-execution" framework. The perception module utilises on-board sensors and high-definition (HD) maps for environmental sensing and localisation; the planning module calculates driving trajectories and decision commands based on this perceptual data; and the control module translates these commands into low-level actuation, including steering, acceleration, and braking.

Technical Routes: Three Mainstream Approaches

There are currently three primary technical routes to implementing AVP:

Vehicle-centric route: The vehicle relies entirely on its on-board sensor suite (LiDAR, cameras, and millimetre-wave radars) for end-to-end perception, decision-making, and control, requiring no modifications to the car park. While this approach was explored early on, it demands high-specification on-board hardware, and relying solely on vehicle-side perception presents inherent limitations in complex car park environments.

Infrastructure-centric route: By deploying sensors such as LiDAR and stereo cameras throughout the car park to handle environmental sensing and parking bay monitoring, the vehicle only requires basic connectivity and drive-by-wire chassis actuation to park under facility guidance. This keeps on-board hardware requirements low, but poses significant challenges due to the heavy capital expenditure needed for large-scale car park retrofitting.

Vehicle-infrastructure cooperative route: The vehicle and car park communicate via the cloud; the facility pre-assigns a parking bay and charts the optimal route, which the vehicle then follows to navigate and park. Leveraging the best of both vehicle and infrastructure capabilities, this is widely regarded as the most commercially viable model today. Chengdu e-Boch Technology employs this approach—data exchange between the vehicle and the car park takes place even before arrival, ensuring far more efficient parking guidance.

Deployment Progress

AVP is currently undergoing a pivotal transition from pilot trials to commercial rollout. The P7 Car Park at Nanjing South Railway Station stands as a benchmark AVP showcase in China—the facility underwent a smart retrofit and integrated an intelligent management platform, embedding a digital twin of the car park into the intelligent driving app ecosystem. Compatible vehicles can autonomously park and be retrieved from any section of the car park, with the initial rollout supporting over 20 models, including AITO.

Over in Chengdu, e-Boch Technology joined forces with Dongfeng Motor to deploy its proprietary AVP technology into mass-production models, rolling out pilot projects at locations such as Xinchuan Innovation Park, Chengdu East Railway Station, and Tianfu International Airport. Statistics show that parking bay utilisation surged by over 30% during the pilot run in the Xinchuan area.

It is worth emphasising that: AVP is classified as an L4 low-speed autonomous driving feature. While it does not require active driver supervision during operation, users are still advised to keep an eye on the vehicle's surrounding camera feeds via their smartphone, remain prepared to pause the manoeuvre if anomalies arise, and ultimately assume full responsibility for vehicular safety.