BETA
This is a BETA experience. You may opt-out by clicking here

More From Forbes

Edit Story

Autonomous Driving Technology At NXP's FTF Appears Closer Than You Expected

Following
This article is more than 7 years old.

With all the hype surrounding autonomous or self-driving cars, you’d think it was right around the corner. Well, you’d be right. The technology to build a self-driving car is here and is being demonstrated at their FTF Technology Forum in Austin this week using components NXP is already shipping.

The talk about autonomous driving cars has centered on it as a luxury item, but in fact, it is a critical safety enhancement that can save many thousands of lives each year. NXP cited a 2015 NHTSA report (NASS National Motor Vehicle Crash Causation Study) that showed 94% of road traffic accidents are caused by driver error or impairment. Many of those accidents can be avoided using autonomous driving technology which can control the car to manage and prevent emergency situations. To get to that level of autonomous control requires additional sensors and lots of processing power. A number of companies are working hard on this problem because the benefits in lives saved and injuries prevented are immense.

NXP’s Autonomous Driving Platform Is Ready to Go

The NXP system demonstration at FTF incorporates the company’s “BlueBox” central computing engine for autonomous vehicles. The BlueBox combines radar, LIDAR (Light Detection And Ranging), and vision sensing, as well as an onboard Vehicle to Everything (V2X) system, to model the vehicle situation and make safety decisions to protect that car and its passengers. All the computing elements use production or sampling NXP silicon, making this system or solution ready for the road.

The company stated that the BlueBox is already in hands at four of the top five largest carmakers in the world. It has been shipping to these select customers since September 2015.  Because NXP products are already qualified by so many manufacturers, the company  is capable of bringing this technology to market faster, giving the company a time-to-market advantage and a wide reach. While there’s plenty of fresh interest in the automotive electronics market these days by chip companies such as Cypress, Intel, Qualcomm, and NVIDIA; after the NXP and Freescale merger, the new NXP is the world’s #1 provider of silicon to the worldwide automotive market. The company also claims to be a world leader in ADAS processors, having shipped more than 30 million ADAS processors worldwide to date.

Autonomous Driving System Architecture

In the autonomous vehicle system, multiple streams of sensor data are routed to the BlueBox engine, where the streams are combined to create a complete 360° situational real-time model of the physical environment around the vehicle. The NXP S32V processor takes the sensor inputs from the platform’s NXP silicon-powered LIDAR, radar and vision nodes, and creates a map via its sensor fusion capabilities. The S32V includes graphics engines, has dedicated high-performing accelerators for image processing and also incorporates automotive-grade functional safety engines.

The BlueBox engine also incorporates the company’s LS2088A embedded compute processor. The LS2088A hosts much of the embedded intelligence and machine learning. The processor integrates eight 64-bit ARM Cortex-A72 cores running at 2 GHz, along with specialized accelerators, high performance communications interfaces and DDR4 memory controllers, making it one of the highest performing embedded products today. The Bluebox can deliver 90,000 DMIPS (million instructions per second) of performance using under 40 Watts of power without the need of exotic thermal management solutions like liquid cooling.

This functionality greatly improves car safety by both managing and preventing emergency situations. BlueBox and its connected solutions also incorporate the embedded intelligence and machine learning required for complete situational assessments, supporting advanced classification tasks, object detection, localization, mapping and vehicle driving decisions.

In addition to performance, NXP has taken a more open-platform approach by using a Linux-based solution that is programmable in C language. By using open sourced and standard programming models, it allows the automotive manufacturers to build their own differentiated products. Each manufacturer will want to craft a different approach to autonomous and assisted control that aligns with the brand image of their cars.

These processors along with other NXP support chips are set to improve auto safety. The BlueBox is a significant platform that auto manufacturers can build on for their customized solutions. As noted earlier, NXP’s BlueBox has already shipped to select customers since September 2015.  Visit http://www.nxp.com/BlueBox for more information.

While BlueBox is a great start for enabling autonomous driving platforms, we’re not done yet. Not even close.  What is still going to take time is autonomous control under extreme weather and road conditions. Self-driving modes have been demonstrated, but when the road is clearly defined and mapped, the visibility and weather is reasonably clear, and the government regulators allow it. Ubiquitous autonomous driving will require additional AI, and compute power to make judgements about snow-covered roads, black ice, heavy fog, dirt roads, major construction detours, etc. In those extreme conditions, there’s still no substitute for the experienced human driver – for now. Eventually, autonomous driving AI should be better than the vast majority of drivers even under these extreme conditions, based on the plethora of sensors, faster sensory integration speeds and faster decision making speed – superhuman capabilities that mere humans do not poses.

Kevin Krewell, Principal Analyst

Tirias Research