Dive Brief:
- Electric automaker Rivian shared details of its autonomous driving technology and new vehicle hardware platform to support it at the company’s “Autonomy & AI Day” event at its Palo Alto, Calif. office on Dec 11.
- The company’s Rivian’s Gen 2 and forthcoming Gen 3 electric vehicles are being outfitted with a more advanced AI-powered software and autonomous driving sensor suite that includes high-resolution cameras, multiple radar units and lidar for the first time.
- According to Rivian, the sensors and computer upgrades, including a new central processor, will dramatically improve its vehicle perception capabilities, which the company says is crucial for safe and reliable hands-free driving.
Dive Insight:
The heart of Rivian’s more advanced hands-free driving system is its new in-house designed processor the company calls the “Rivian Autonomy Processor” (RAP1). The RAP1 will power the company’s third-gen “Autonomy Compute Module 3” (ACM3), which can process 5 billion pixels of sensor data per second, according to Rivian.
The RAP1 also features low latency interconnect technology, allowing additional chips to be connected to multiply total processing power, according to Rivian. The immense processing power of the ACM3 will support the company’s goal of launching more advanced SAE Level 4 automated driving capabilities in the future that can operate without human oversight.
Rivian also shared details of its updated “software-first approach for self-driving”, which is powered by the company’s new “Rivian Autonomy Platform” using an end-to-end data loop for training. It allows the system to improve over time using data collected from Rivian’s fleet. This approach, often referred to as a “data flywheel”, is typically used to handle challenging edge cases, which are rare scenarios a self-driving vehicle may encounter that human-coded software cannot handle.
“Our approach to building self driving is really designed around this data flywheel, where a deployed fleet has a carefully designed data policy that allows us to identify important and interesting events that we can use to train our large model offline,” Rivian CEO RJ Scaringe said during a presentation.
The company also announced that software advancements will be rolling out to its second generation R1 vehicles in the near term, including the addition of “Universal Hands-Free” (UHF) assisted driving for extended periods and to more roads. It will be available on over 3.5 million miles of roadways in the U.S. and Canada, up from the previous 150,000 miles.
Among the other highlights of Rivian's AI-powered autonomous driving platform is a Large Driving Model (LDM) trained on massive datasets. It utilizes what’s called Group-Relative Policy Optimization (GRPO) to compare multiple driving trajectories. For example, if a Rivian vehicle operating in autonomous mode is approaching a busy intersection, the system will analyze multiple paths forward and choose the best one, which is then added to the model.
When combined with its multi-modal sensor strategy using cameras and radars, it will improve real-time detection of the area around the vehicle, according to Rivian.
Rivian also announced a new autonomous subscription plan called “Autonomy+”, with continuously expanding capabilities to get better over time. It’s slated to launch in early 2026 and will be priced at $2,500 (one-time) or $49.99 (per month).
Rivian said its Gen 3 Autonomy hardware, including the new ACM3 processor and lidar, is currently undergoing validation. The company expects the hardware to be made available for the R2 SUV starting at the end of 2026.
“We are very proud to be at the leading edge of multi-modality sensing and to be continuing our trajectory of vertical integration with our RAP1 chip and Gen three autonomy computer,” said Vidya Rajagopalan, Rivian’s SVP of electrical hardware, in a presentation. “We expect that at launch in late 2026 this will be the most powerful combination of sensors and inference compute in consumer vehicles in North America.”
