Here’s a challenge: try to find a review of the Toyota Corolla that doesn’t bemoan its numb steering. Now try with a Chevy Cobalt. Or a Venza, or Vibe, Or Rav4, or Equinox. What do these vehicles have in common? Column-mounted electric power steering systems from JTEKT, a Toyota spin-off supplier which has done a brisk business in these fun-eliminating steering systems. And though the motor press has been bashing electric power assist steering (EPAS or EPS) for its deleterious effect on handling, the explosive growth in these systems may put more at risk than mere enthusiast-approved steering feel.
This anesthetization of steering systems has not taken place because manufacturers appreciate the proliferation of words like “numb” and “overboosted” in reviews of their products. EPS offers improved efficiency due to its reduction of parasitic losses, and is cheaper to manufacture than traditional hydraulic systems. This killer combination offers manufacturers a combination of improvements that have proven near-impossible to resist, resulting in the broad proliferation of EPS systems. And if reduced steering feel were the only casualty of the switch, it would be a tradeoff that any manufacturer would be willing to run.
But as EPS has exploded onto the market, a number of troubling issues has plagued the system. The National Highway Traffic Safety Administration has opened investigations into the Chevy Cobalt and Toyota Corolla, which share the column-mounted JTEKT EPS system. Cobalt, which moved to an EPS system for the 2005 model year has been haunted by an accelerating number of failures since the switch, while the Corolla investigation centers on Corollas built since the 2009 model-year switch to EPS.
In both of these vehicles, pinning down exact steering defects is proving to be difficult. Many of the Corolla complaints are related to sudden veering, particularly at speeds above 40 mph. The Cobalt, meanwhile, seems to experience complete EPS failure, causing momentary loss of steering and/or the need for drastically increased steering effort. Again, the inability to stay in a set traffic lane is being targeted as the most dangerous symptom of the possible defect.
These symptoms fit conveniently into a category that an early report (by Amit Rohidas Bendale of the Vishwakarma Institute of Technology, publication date unknown) on EPAS technology posted at college-seminars.com calls “auto steer,” a term the paper asserts “has crept into the lexicon as an adjunct to the development of EPAS system.” Bendale attempts to explain this phenomenon in his paper’s section on the disadvantages of EPS thusly:
To-date, technical and product liability concerns have precluded the introduction of such systems in the U.S. market through it is expected that niche application may be expected in the near-to-mid term mix of future vehicles. Such system design have yet to prove themselves sufficiently reliable and safe to prevent dangerous “auto steer” events. “Auto steer” has crept into the lexicon as an adjunct to the development of EPAS system. As the name implies “auto steer” denotes an uncontrolled steering event neither commanded nor stoppable by the vehicle’s driver due to catastrophic failure in the electron hardware or software. In truth, these systems are control servo systems, similar in function to aircraft control servo systems, and must have multiple redundancy. Although these new EPAS systems are said to have multiple redundancy , their design and broad application within the automotive industry have been, and will continue to be, subject to economic pressure more extreme then found in the aircraft industry. For instance one obvious safety related item has been universally deleted from such system specifications: a clutch for physically disengaging the reduction gear box and drive motor assist assembly from the host steering system in the event of system failure. This means that a driver encountering an EPAS system failure will have to exert additional force to “back drive” (i.e. manually over ride) the systems reduction gear box and drive motor assist assembly while attempting to maintain control of the vehicle in the absence of normal power steering assist.
Already, the parallels between these EPS issues (which are admittedly theoretical, but reflect NHTSA complaints) and Toyota’s gas pedal issues are plain to see. The first is the issue of economic pressure towards cost-cutting, which leads to the reduction of fail-safes and redundancy. Another parallel is the fragility of by-wire control systems: in Toyota gas pedals, tiny amounts of moisture was enough to cause the pedals to stick. In the case of EPS steering, there’s evidence to suggest that even cell phone interference could cause system failure. This is not to say that these two situations are directly comparable, but both cases indicate that earlier mechanical systems offer few clues as to possible malfunctions of electronic systems. Finally, the most important parallel between Toyota’s “unintended acceleration” and EPS “unintended veering” is the complicated dynamic between the driver and the system at the moment of malfunction or failure. Bendale continues:
Unlike the manual system described above, PAS with the presence of supplementary steering force to that provided solely by the operator introduces additional engineering challenges in terms of maintaining the desired steering linearity described previously. In fact, with respect to steering linearity, a poorly designed power steering assist system may have almost no relationship between the hand wheel torque applied by the operator and the actual required steering force imposed by the wheel or tires. There no longer may exist the uniform, consistent and predictable relationship between the “input and outputs” to facilitate “tactile reference driving.” Restated, the tactile sense of the driver to maintain directional control, and the magnitude and modulation of the “input” force may not bear a direct, proportional relationship to the required “output” force delivered by the steering system. Tactile reference steering is simply not possible with such vehicles. Rather their drivers must continuously engage in “visual reference steering” to maintain directional control. The result is that such vehicles are very tiring to drive for any length of time or distance. Further their drivers are constrained to continuously look at the road. If such driver should look away even momentarily (i.e., to check a rearview mirror or a child in the car), he or she has minimal tactile reference as to the actual position of the vehicle during that period of time. This is dangerous because, depending upon the road topography and condition, the vehicle may have moved transversely in significant amount relative to where the driver thought his or her vehicle was positioned. This can and often does lead to serious trouble.
This helps explain the varying accounts of unintended veering, as the ability to handle sudden changes in steering response varies greatly from driver to driver. Like unintended acceleration, this makes the task of narrowing down the primary causes of EPS failure or malfunction impossible based on recorded accounts alone.
More importantly, this helps explain the link between numb, overboosted and uncommunicative steering and safety. At highway speeds, any failure or malfunction would be registered and reacted to by touch (tactile feedback) before the brain would be able to register visual cues that vehicle control has been compromised. Because of EPS’s deficits in tactile feedback, and consistency between wheel and steering wheel positions, that all-important emergency tactile feedback would be more difficult for even a well-trained driver to interpret, making a crash more likely even if the unassisted steering system were still functioning (as has been reported in certain Cobalt incidents).
This is clearly an issue that companies like JTEKT have known about, as a 2008 paper on EPS [PDF] by a JTEKT engineer admits that making EPS feel like the hydraulic systems the buying public is used to has been a major challenge. Another paper from the JTEKT Engineering Journal [PDF] indicates that on higher-power versions of column-mounted EPS,
sensitivity to such inverse inputs as flutter and brake vibration has become higher. J-ISM [JTEKT's EPS assist control algorithm] has adopted a suppression system that can detect inverse input vibration as torque differentiation value, and by providing assist in the direction of canceling the vibration, the vibration is not transmitted to the steering wheel.
In short, tactile feedback is actively canceled, not only reducing the driver’s ability to respond to an emergency based on steering feel, but also further loading the EPS system with compensation duties in addition to pure steering control.
The point of all this is not to merely raise unwarranted fears about EPS. Improved efficiency and lower costs are the very definition of a successful automotive system, and as long as EPS offers these benefits, its use will continue to grow. As this trend continues, it will be important for safety watchdogs to pay close attention to these systems. Without pressure to include redundancy, fail-safes and extremely robust sensor units, suppliers will continue to cut the cost of these systems to remain competitive with the growing number of firms that offer similar systems. Just as importantly, these new electronic systems require a new approach to safety that pays as much attention to driver reaction to malfunctions as to the malfunctions themselves. Only in this way can we be sure that relatively minor malfunctions don’t yield an epidemic of inexplicable accidents and recalls.