DiDi Autonomous Driving updates cooperation fruits with GAC AION

Shanghai (Gasgoo)- On April 12, the 2025 GAC TECH DAY and the GAC Intelligent Safety Summit took place in Guangzhou, with DiDi Autonomous Driving attending as a key strategic partner.

At the event, DiDi co-founder and CEO of DiDi Autonomous Driving, Zhang Bo, shared the company's latest developments and perspectives in L4 autonomous driving, while also unveiling its next-generation hardware platform for factory-installed mass-produced Robotaxis.

Another highlight of the event was the debut of the first L4 production model jointly developed by DiDi Autonomous Driving and GAC AION. The vehicle integrates DiDi's newly upgraded hardware architecture and comprehensive safety redundancy system with GAC AION's global-ready platform available for L4 autonomous driving, offering robust adaptability for worldwide deployment. Mass production and deliveries are targeted for the end of 2025.

The collaboration builds on the "AIDI" initiative launched in May 2023, with both companies pooling their technological strengths to bring L4 self-driving capabilities to market.

According to Zhang Bo, the upcoming production vehicle features DiDi Autonomous Driving's latest hardware platform, equipped with 33 sensors in total. This multi-sensor fusion system includes LiDAR, cameras, 4D millimeter-wave radar, infrared cameras, and acoustic sensors, enabling full 360-degree perception across all driving conditions.

Enhanced by a new LiDAR chipset, the 4D radar delivers four times the performance of its predecessor, improving performance in challenging weather such as rain, snow, and fog. Meanwhile, the infrared camera boosts pedestrian detection in low-light environments, and DiDi has also become the industry's first to bring solid-state blind-spot LiDAR into volume production.

Powering the system is DiDi's fully self-developed Orca central computing platform, billed as the industry's first mass-produced, tri-domain integrated vehicle brain. It combines the intelligent driving, cockpit, and communication-navigation domains into one unified architecture. The Orca offers GPU computing power exceeding 2,000 TOPS and a 48-core CPU, with sensor connectivity capabilities doubled from the previous generation.

The event also marked the first public reveal of the sensor layout on the new L4 vehicle model. The car features a futuristic array including a top-mounted floating disc-style LiDAR, radar units embedded in the front fenders, and front & rear-mounted cameras—all arranged for panoramic 360-degree coverage. A full-width light bar integrates the front headlights with an interactive display, capable of showing greetings, messages, emojis, and real-time vehicle status, enhancing both user engagement and recognizability.

In terms of safety, the vehicle boasts a multilayered redundancy architecture across algorithms, software, and hardware. Algorithm redundancy includes an independent collision risk detection module, while the software layer features multi-scenario minimum risk condition (MRC) strategies to define appropriate responses to different failure levels via L1 Fallback. Hardware redundancy enables L2 Fallback for safe roadside or in-lane stops in the event of component failure.