ADAS & AD

Hailo AI Processor for ADAS & AD

AI-Powered Driving Automation

Since the invention of the first car, safety has been the driver for innovation and progress in the automotive industry. These days, machine learning is harnessed to further drive safety in automotive, but also to enable efficiency and convenience for drivers. Machine learning is applied in both Advanced Driver Assistance Systems (ADAS) and in autonomous driving (AD), which rely on AI algorithms to analyze and interpret data from cameras and other perception sources, like radars and LiDARs to navigate roads and make driving decisions.

These enhancements also find their ways to lightweight vehicles, motorcycles and bicycles, as well as off-road vehicles, giving rise to different sets of applications that go beyond driving on a motorway.

Driving Automation on the Edge

From lane keeping to traffic jam assistת to highway pilot, and eventually to urban pilot – a compute-intense AI-focused machine is needed, which needs to process input from multiple sensors to create a reliable 360-degree situational awareness and bird’s eye view image of the road in real-time. The system must also meet multiple constraints including thermal dissipation, mass production quality, size and price. ​​

Hailo is working with leading OEMs and Tier-1s who are at the forefront of AI automotive technology, to introduce state-of-the-art leading deep learning performance into various assisted and autonomous driving systems, while meeting evolving safety standards and regulations. ​The Hailo AI processor is an AEC-Q100 grade 2 qualified Automotive AI accelerator, which has been developed in accordance with ISO26262 guidelines to meet ASIL-B conformance.

Benefits

High AI Compute Power

Hailo’s dedicated AI processors provide the high power-efficiency (high processing throughput per unit of power) that allows to pack more AI processing into the vehicle, making it possible to scale the AI compute to the required level of autonomy

Thermal Efficiency

The Hailo processors enable high AI compute power at an exceptional power consumption of as low as 2.5 W, eliminating the need for active cooling

Cost Efficiency

The Hailo AI processors are specifically designed for processing NN models at a very efficient usage of the silicon area, resulting in low cost per processing unit ($/FPS). As the Hailo processor does not typically require external memory and consumes a very low amount of power, significant cost reduction can be achieved on RBoM and thermal solution for heat dissipation

Openness

Open and agnostic software ecosystem, allowing OEMs and Tier1s control and innovation in their ADAS and AD applications

Proposed Solution Outline

Hailo offers maximum flexibility and AI compute scalability in ADAS/AD ECU design, by combining automotive SoCs with the Hailo AI processor or multiple processors, depending on the number of sensors and complexity of models.

One small, low-power Hailo chip is able to process multiple video streams and multiple chips can work in tandem or cascade, all while maintaining industry-leading efficiency and scalability, high processing throughput and low latency. The solution’s flexibility allows the use of different types of camera inputs, ranging in resolution and input size.​

Explore more

Demo
Bird’s Eye View 3D Perception Solution
Demo
L2+/L3 Full Surround Perception for ADAS and Automated Driving

ADAS

Making Attractive ADAS Applications Cost-Effective

Advanced Driver Assistance Systems (ADAS) are continuously evolving to offer drivers a safer and more comfortable journey. These systems combine processing of different sensor modalities with real-time data processing for enhancing both driving and parking capabilities.

AI has a pivotal role in ADAS, since it is in the core of the vehicle’s perception of its environment. Traditional ADAS processors have suffered from lack of AI performance and/or from inadequate power consumption to meet automotive ADAS ECU requirements. Since the AI model landscape has changed significantly in recent years, many processors have also fallen short in their ability to run state-of-the-art Neural Networks (NNs), requiring alternative solutions.

The Hailo AI processors are designed for scalability and can support the demanding deep learning workloads that ADAS require. One small, low-power chip can process multiple video streams and multiple chips can work in tandem or cascade, all while maintaining industry-leading efficiency and scalability, high processing throughput and low latency. The solution’s flexibility allows running state-of-the-art Neural Networks including transformers, alongside the use of different types of sensor inputs, ranging in resolution and input size.​

Explore more

Blog
Backing into the Future: Unlocking the Potential of Automated Parking
The technology advancements and market drivers that accelerate the transition to automated parking
Blog
ADAS Anatomy: from SAE to NCAP
While my experience wandering the halls of CES 2020 more than a year ago seems as nostalgic as far more distant memories after this unusual year, I clearly remember my impression on automotive ADAS trends at the time. A year later, wandering the virtual halls of CES 2021, I had a similar impression, even though…
Blog
Leveraging Vendor Partnerships for ADAS Success: LeddarTech and Hailo
It’s a late summer evening, you’ve had a long day at work and all you want to do is get home and relax, but the usual horrible traffic jam is worrying you. The thought of spending the next thirty minutes switching between the accelerator and brake pedals is frustrating. As per the INRIX 2022 report,…

Highly scalable solution for AI-based fusion & perception in all classes

Autonomous Driving (AD) is a technology that enables a vehicle to operate without direct human input. It refers to a vehicle’s ability to drive itself without human intervention, including steering, accelerating, and braking. The software and AI algorithms used in AD systems process data from multiple sensors and make decisions about how the vehicle should respond to its environment. This may include adjusting speed, changing lanes, avoiding obstacles, and more.

The higher the level of autonomy, the more complex and advanced algorithms are needed, and the more AI capacity is needed to analyze the huge amount of data from multiple sensors, and take safe and efficient driving or halting decisions in real-time. For this reason, the role of AI processors in autonomous driving is increasingly prominent.

While there is still much development and testing required before fully autonomous passenger vehicles can be available on the market, AD technology has the potential to revolutionize transportation by increasing safety, improving traffic flow, and significantly enhancing humanity’s productivity. This technology is making its way to a variety of markets such as robotics, heavy machinery, offroad vehicles, agriculture and more, in which Hailo is actively working with customers and partners to automate daily operations.

Explore more

Presentations
AutoSens Interview with Yaniv Sulkes, VP of Automotive
Blog
Pairing Sensing with AI for Efficient ADAS
Sensor and sensing capabilities are common and expanding in modern vehicles. One of the major motivators for this is safety or, more specifically, Euro NCAP Vision Zero guidelines. These are driving automakers’ and Tier 1 suppliers’ ADAS/AV roadmaps, requiring more powerful scene understanding. Among other things, extending the scope of VRU (vulnerable road users) protection…
Webinar
Scaling AI in Automotive with Hailo AI Acceleration
Scaling AI in Automotive with Hailo AI Acceleration

Breathe life into your edge applications with the Hailo AI processors