Convergence in Self-Driving Vehicles

By Contributed Article | March 31, 2014

The latest automotive technology is sensor-based, but to make self-driving vehicles a reality, technology convergence is necessary. This article, based on research by KPMG, sheds light on what may be next in automotive connectors and other technology.

This article is excerpted from a white paper published by KPMG.


Can we build a safe, self-driving vehicle? Yes. In fact, Google has already logged more than 200,000 miles in a fleet of self-driving cars retrofitted with sensors. And Google is not alone; traditional automakers and suppliers have also developed self-driving functionality that uses sensor-based solutions and they have a host of new applications in the pipeline. At the same time, a number of organizations, including automotive and high-tech companies and the USDOT, have been focused on the potential for using connected-vehicle communication technologies for collision avoidance and traffic management.

What’s missing, so far, is the convergence of sensor-based technologies and connected-vehicle communications that is needed to enable truly autonomous vehicles. Let’s discuss the existing technologies, their current limitations, and why we believe they are likely to converge in the not-so-distant future.

Sensor-Based Solutions

The automotive industry is currently developing sensor-based solutions to increase vehicle safety in speed zones where driver error is most common — at lower speeds, when the driver is stuck in traffic, and at higher speeds, when the driver is cruising on a long stretch of highway. These systems, known as Advanced Driver Assist Systems (ADAS), use a combination of advanced sensors, such as stereo cameras and long- and short-range RADAR, combined with actuators, control units, and integrating software, to enable cars to monitor and respond to their surroundings. Some ADAS solutions, such as lane-keeping and warning systems, adaptive cruise control, back-up alerts, and parking assistance, are available now. Many others are in the pipeline.

The next generation of driver-assist systems will likely offer greater vehicle autonomy at lower speeds and may reduce the incidence of low-impact crashes.

Companies are also developing sensor-based, driver-assisted solutions, which use stereo cameras and software and complex algorithms to compute, in real time, the three-dimensional geometry of any situation in front of a vehicle from the images it sees.

Such sensor-based systems offer varying degrees of assistance to the driver, but, in their current form, are not yet capable of providing self-driving experiences that are complete and cost-competitive. Their limitations include:

  • Perception of the external environment: So far, the fusion of available sensors and artificial intelligence is not capable of “seeing” and understanding the vehicle’s surroundings as accurately as a human being can. Humans use a combination of stored memories and sensory input to interpret events as they occur and anticipate likely scenarios. For example, if a ball were to roll onto a road, a human might expect that a child could follow. Artificial intelligence cannot yet provide that level of inferential thinking, nor can it communicate in real time with the environment. “These algorithms are very complex and will need to replace over 16 years of human learning,” explained Christian Schumacher, head of systems and technology for Continental Automotive Systems, N.A.
  • Cost: Creating a 360-degree view of the vehicle’s environment requires a combination of sensors and may cost more than consumers are willing to pay. Light Detection and Ranging (LIDAR)-based systems provide 360-degree imaging but are complex, expensive, and not yet ready for the market. The LIDAR system used in the Google car, for example, costs $70,000. Value-chain stakeholders will need to have a clear and compelling business case before investing in this technology.

Connectivity-Based Solutions

Connected-vehicle systems use wireless technologies to communicate in real time from vehicle to vehicle (V2V) and from vehicle to infrastructure (V2I), and vice versa. According to the USDOT, as many as 80% of all crashes — excluding those in which the driver is impaired — could be mitigated using connected-vehicle technology.

Dedicated Short-Range Communication (DSRC), which uses radio waves, is currently the leading wireless medium for V2V communication. It operates at 5.9GHz frequency, using standards such as SAE J2735 and the IEEE 1609 suite (protocols that establish what messages are sent, what the messages mean, and how they are structured) and is being tested rigorously to see if it can fully support V2V cooperative safety applications. Currently, DSRC offers the greatest promise, because it is the only short-range wireless alternative that provides all of the following:

  • Fast network acquisition
  • Low latency
  • High reliability
  • Priority for safety applications
  • Interoperability
  • Security and privacy

These features are especially important for active safety applications, because safety-critical communication must be reliable, immediate, network and device “agnostic,” and secure. Another benefit of DSRC is that it operates using free spectrum, which is already reserved by the US government for transportation applications.

Within the automotive industry, two entities have emerged for testing and developing V2V and V2I communications. The Vehicle Infrastructure Integration Coalition (VII-C) is a collaboration of federal and state departments of transportation and automobile manufacturers. In 2009, the coalition published the results of its connected vehicle concept testing; it is now focused on policy issues that must be resolved before the technology can be deployed. Another group, the Crash Avoidance Metrics Partnership (CAMP), held driver clinics in six US locations as part of a Connected Safety Pilot.

To move beyond the test phase and set the stage for self-driving vehicles, a number of obstacles must be overcome:

  • Critical mass: Because V2V communication requires other similarly equipped vehicles for sending and receiving signals, the technology will not achieve its potential until the capability is ubiquitous. That may require mandates and will certainly require cost-effective solutions and the ability to retrofit existing vehicles.
  • Infrastructure modifications: V2I communication for active safety will require infrastructure equipped with DSRC-compliant transceivers, and the cost of building that infrastructure may present barriers. An intermediate solution might focus only on crash avoidance at high-volume or other critical intersections. Another solution could use cellular technology and its existing infrastructure for longer-range communication and DSRC for shorter ranges. Heri Rakouth, manager of technology exploration at Delphi, noted: “Advances in cellular technology could be a longer-term solution to the infrastructure investment cost that is associated with DSRC.” However, some inherent shortcomings exist with cellular technology for use in active safety systems: It suffers from latency issues (it is too slow) and has bandwidth constraints, both of which reduce its viability for safety-critical applications.
  • Dependency on sensors: Although connected vehicle solutions can communicate with the external environment, sensor-based solutions will need to co-exist in order to cover situations that involve obstacles — obstructions in the road or pedestrians, for example — that would not be connected to and communicating with the network.

The Benefits of Convergence

The convergence of communication- and sensor-based technologies could deliver better safety, mobility, and self-driving capability than either approach could deliver on its own. As Pri Mudalige, staff researcher for General Motors’ Global R&D, said, “V2V technology…may simplify the all-sensor-based automotive advanced driver-assist systems, enhance their performance, and make them more cost-effective.”

Indeed, our list of the top benefits of convergence corresponds with Mudalige’s and includes:

Timing and cost: Convergence would help reduce the cost and complexity of standalone solutions. Adding DSRC would eliminate the need for the more-expensive sensors and bring down the cost of the overall package.

Proxy for human senses: Convergence would increase the inputs that are available for decision-making and reduce the need for more sophisticated artificial intelligence. The combination of sensors and connected-vehicle solutions would allow self-driving vehicles to collect the requisite information to make real-time “decisions” and respond to the myriad on-road scenarios drivers face every day. Whereas sensors can see what is directly within their frame of vision, V2V communication adds the potential for trajectory prediction, as vehicles communicate their intentions to each other, lessening the reliance on artificial intelligence.

Functionality redundancy: There is no room for error with safety-critical functionality. The technology has to work 100% of the time; the combination of connected vehicle technologies and sensor solutions would provide a necessary level of redundancy.

Infrastructure investment: Connected vehicle solutions require large-scale infrastructure investments. Convergence could help mitigate some of this requisite investment by covering some use cases using sensors.

The Path to Convergence

There are still significant hurdles on the path to convergence.

Improved Positioning Technology

GPS offers some promise, but the technology, which pinpoints location within +/- 10 meters, isn’t accurate enough to be used for safety-critical applications. GPS error-correction technologies, such as RTK (real-time kinematics), are expected to be introduced in the future as the demand for accurate positioning increases and cost curves permit mass-market introduction.

High-Resolution Mapping

Today’s digital maps lack the necessary detail to support self-driving applications, which need to “see” the environment in as much detail as the human eye. If a firm can successfully resolve the accuracy issue, it would help alleviate some infrastructure burden of a DSRC-only solution.

Reliable and Intuitive Human/Machine Interface (HMI)

The interface between driver and machine remains a complex problem. Drivers must know when and how to hand off control and take it back. That handoff must happen seamlessly, instantaneously, and safely — and drivers must be thoroughly comfortable with the process in any vehicle they use.


The regime for connected vehicles is fairly mature based on the SAE J2735 and IEEE 1609 standards, but additional standards will be needed to ensure full interoperability. A mandate, if it occurs, should provide momentum to develop them, but a question remains: What gets standardized, and what remains part of the branded experience controlled by manufacturers?

For decades, KPMG’s Global Automotive practice has been recognized for its commitment to the industry. Through its international network of member firms, it has the global reach and experience to serve clients anywhere in the world. To read this white paper in its entirety, click here.

Get the Latest News