Suppliers are currently developing a fusion of sensors that will create “an electronic cocoon” around vehicles that will provide the information necessary to mitigate collisions. This illustration from TRW shows how future cars will bristle with sensing capabilities.
Adaptive cruise control (ACC), which uses long-range radar to sense cars ahead and automatically brake if they slow down, is the first step in creating a sensor web for vehicles, and it is already on the road. The Volkswagen Phaeton features an ACC system developed by TRW; the BMW 7-Series is equipped with one from Bosch; and the 2003 Jaguar XKR just debuted with Delphi’s “Forewarn” system.
Though the proliferation of sensors on vehicle chassis have arguably given them the sense of touch, the eyes and ears of the driver are still the primary source of information needed to avoid collisions. That’s about to change. Chassis technology suppliers like TRW Automotive, Delphi Automotive and Robert Bosch Corp. are busily developing and commercializing a diverse array of sensors that will work together to allow vehicles to “see” objects and people and react appropriately. The potential payoff in reduced deaths and injuries is big. How big? “Collectively, these technologies are capable of reducing automobile crashes by an order of magnitude,” says Dave McLellan, industry consultant and former Chevrolet Corvette chief engineer, which could fundamentally change the safety debate.
But that kind of dramatic improvement is a few years off. The consensus among system suppliers seems to be that collision mitigation capabilities gradually will be added to vehicles over the next 10 years or so, beginning, as expected, with higher end cars and SUVs. Some of the sensor infrastructure that will be necessary is already in place in the form of stability control and anti-lock braking systems, and more will come as electronically actuated X-by-wire components filter into the marketplace.
The first real testing ground for these technologies is adaptive cruise control (ACC). Current versions use a 76-GHz narrow-beam radar to detect moving objects directly in front of a vehicle and adjust vehicle speed accordingly. But ACC is not designed to work at slower speeds (lower limits vary between 20-35 mph) and it is blind to stopped objects. Developers have chosen to set these limits for early systems because it both simplifies and reduces component costs, and results in fewer false alarm warnings to drivers. However, future generations will incorporate “follow-stop” capabilities that will brake to a full stop based on the behavior of the preceding vehicle. And eventually “stop-and-go” will be achieved, which will let the vehicle automatically accelerate after a full stop.
With these new capabilities will come increased risks and complexity. In addition to the long-range radar unit of current systems, future ACC units will have to employ short-range 24 GHz radar or video sensors that will provide full coverage across the front of the vehicle and create what Bob Rivard, vice-president of advanced technology development, Robert Bosch, calls “an electronic cocoon.” To do this in a way that will ensure pedestrian safety, without triggering an egregious number of false alarms will require improvements in software beyond what exists today. “Algorithm development needs to continue to advance and become more and more complex and robust, “ says Paul Martindale, product line planner for warning systems at Delphi. But as that happens, new areas for improving safety open up. For example, more sophisticated path determination algorithms fed by information from vision systems will extend the distance at which potential collisions can be identified, thus effectively offering a margin of safety at higher speeds.
Radar and video. A consensus seems to be forming that a fusion of radar and video sensors will ultimately make up the electronic cocoon. Though laser-based sensors are currently cheaper than radar, they are far more susceptible to things like rain, snow and fog, which may ultimately prove their downfall. However, radar still must climb a significant development curve. According to Martindale, “Miniaturizing and mass-producing radar is a challenge. It was developed from military aircraft, where neither cost nor size were big concerns.” And where suppliers weren’t squeezed by their customer for every penny.
Video is less of a problem from the packaging viewpoint because the dime-sized sensors are easily integrated into a vehicle, and current raw video capability is sufficient for its main tasks: the gross recognition of people and objects for collision avoidance, and the “reading” of lane markings needed to alert drivers when they are inadvertently crossing into an adjacent lane. But dynamic range needs to be improved, since bright lights can temporarily blind the sensors, and the amount of processing power required for even simple video images is enormous and expensive. However, the sensor fusion approach can help to mitigate the shortcomings of each type of sensor by having them work together to create a whole that is greater than the sum of its parts. For example, one way to reduce the amount of video processing power needed is to use radar for the initial identification of objects, allowing the video sensors to focus on a much smaller area.
Authority Figuring. Technology developers reckon that all of the building blocks necessary for collision mitigation will be in place in around a decade. In addition to a sophisticated sensor net, these would include: electric power steering, electric power brakes with independent control of each wheel, advanced yaw control and rollover reduction systems. There is a strong belief that future systems will be increasingly robust and reliable, and capable of greatly enhancing safety. That is, if automakers let them. How much authority automatic systems should be given over the control of vehicles is a big question. Everyone’s worst nightmare is a public perception of a Robocar that circumvents driver control. “I think we are going to see a very gradual transition to being allowed by the vehicle manufacturers to give the system more and more authority for things like braking,” says Craig Tieman, product marketing manager for Chassis and Safety Electronics at Delphi. To which McLellan adds: “Collision avoidance systems should intervene as a last resort to avoid a collision, then return control to the driver.”
In fact, suppliers stress that systems indeed will be designed to warn drivers of potentially dangerous situations, rather than immediately take control. But in some situations, they would be programmed to initiate actions independently. A mild example of this is a lane-keeping function that first warns a driver that he is veering into another lane, then applies a minute torque overlay that tweaks the steering to keep the car in its lane. A more serious scenario might involve an impending crash with a speeding vehicle. If the sensor net determined that a collision was imminent, the system could brake the vehicle, pre-charge the airbags, pre-tension the seatbelts, and plot a path to impact that would result in the least likelihood of injury or death.
Still, even with near-perfect computer calculations and the requisite scope of control, these systems will not be a panacea. Phil Cunningham, director of Product Planning, Chassis Systems, at TRW Automotive offers this perspective, “My vision is that in 10 years time we should be able to control everything about the chassis that is feasible within the laws of physics. But I don’t know if we will ever see true collision avoidance. What we will have is collisionmitigation.”
“It’s all about cost.” That’s Delphi’s Tieman on the biggest stumbling block for getting collision mitigation systems into cars sooner. He says, “The technologies are readily available, performance is quite good–it’s a matter of being able to get them into a mass production mode at an attractive cost.” TRW’s Cunningham echoes the sentiment, “If the vehicle manufacturers were a bit more aggressive in terms of the fitment rate we would see the costs of this technology come down drastically.” But suppliers are not relying entirely on increased volumes to bring costs down. Cunningham points out that sharing the sensors needed by different systems can eliminate duplication and greatly reduce purchasing costs. He says TRW is already doing this by sharing a single steering angle sensor between its electric power steering and traction control systems. And posits that even such seemingly disparate functions as rain-sensing wipers and lane keeping could use the same video sensor.
Consultant McLellan, however, takes a different approach to the cost issue. He thinks that once collision mitigation systems are in place and accident rates drop, car owners will save as much as $1,000 a year in premiums. “These technologies will ultimately pay their own way,” he predicts, “and customer demand will drive the availability of these systems across the fleet.”