Volvo ES90: A New Chapter in Automotive Safety
Volvo, long synonymous with safety, has launched the ES90, an all-electric vehicle aiming to maintain its leadership in the age of digital technology. The ES90 is more than just a car; it’s a statement of the company’s commitment to redefining safety through advanced computing and software.
Alwin Bakkenes, Head of Software Engineering at Volvo, discussed the ES90 and its implications for the brand’s core objectives. Volvo’s shift focuses on software as a driving force in innovation, prioritizing human-centric technology and advanced safety systems.
Premium EV Features and Design
Although not yet available for test drives, the ES90’s design is promising. It has a hatchback design that resembles a sedan, providing practicality for families. It also has spacious interiors, with ample room for adults, as well as a long electric range (up to 435 miles in the WLTP cycle).
The vehicle’s 800V architecture enables rapid charging, and battery sizes up to 106kWh, alongside a dual-motor option, signal the ES90’s premium positioning.
The Power of Computing
The ES90 stands out due to its computing capabilities, solidifying its place as Volvo’s flagship. The vehicle is powered by two NVIDIA DRIVE AGS Orin processors, with a combined computational power of about 500 trillion operations per second (TOPS). This makes it the most powerful Volvo car ever built. These chips are designed for AI inference tasks which are a key component of Volvo’s safety stack.
Volvo’s Superset tech stack underpins the ES90, enabling over-the-air updates. Interestingly, the EX90, which uses the same SPA2 underpinnings, initially had a single Orin chip. Volvo plans to offer a free upgrade to existing EX90 owners, adding a second Orin CPU. The Orin’s higher processing power allows for significant enhancements to the AI models that assist in the vehicle’s safety features.
Advancing Software-Enhanced Safety
Even though the ES90 shares its sensor array with the EX90, including cameras, radar, and LiDAR, the enhanced compute performance significantly boosts its software-enhanced safety. The increased processing power allows for more edge inference. This improves system performance, helping the car ‘look out for you’ with an excellent perception stack. This stack utilizes cameras, radars, and LiDAR to maintain reliability in varied conditions.
LiDAR’s long-range, dark-condition obstacle detection is combined with an advanced driver understanding system. This technology interprets driver attentiveness. This creates a holistic Advanced Driver-Assistance System (ADAS) solution.
A Commitment to Real-Life Safety
Volvo aims to deliver technology that goes beyond the minimum requirements for a top safety rating. The goal is to optimize for real-life safety scenarios, with improvements delivered through software updates. Data collected from the vehicles via over-the-air updates allows Volvo to analyze and improve its systems continuously.
The ES90’s computing power contributes to this optimization process, and also benefits from it. The Superset tech stack enables the collection of raw sensor data used to train and improve AI models. These models are then deployed back to the vehicles, enhancing the ADAS stack for not only the ES90, but also the EX90 and future models built using the Superset tech stack.
Bakkenes notes that the ES90 is designed to evolve continuously, leveraging its compute power and sensor array to keep occupants and their surroundings safe. This approach is referred to as Safe Space Technology.
Technology for Future Autonomy
“Collision avoidance is one of the most important things, says Bakkenes, adding that the ES90 and EX90 are well-equipped for autonomous capabilities. Redundancy is a key feature, with redundant braking, steering, and power supply built into the vehicle’s design.
Volvo is rolling out supervised autonomy first, and preparing the hardware for unsupervised autonomy at a later date. The addition of a second Orin chip is based on the computing requirements needed for self-driving features. The 500 TOPS enables edge inference. The vehicles have 5G connectivity to send perception data to a data center, where it is processed and analyzed to improve the training stack.
“It’s a double revolution,” says Bakkenes. ” The car is a computer now, but you’ve also got the ability to fast track development by going through many scenarios in simulation with a digital twin before testing them physically.”
Volvo leverages a Superset tech stack that can be deployed across multiple models to enable functionality to be inherited across vehicles.
Volvo, a part of the Geely Group, will maintain a focus on safety as a unique feature. The computing and sensor technology enable a customer experience built for ultimate safety.
Bakkenes notes that Volvo has been data-driven since the 1970s and has amassed data related to crashes and near-crashes since 2020 to enhance safety. This information is used to fine tune the ES90’s advanced safety features.
