Haptic Vehicle Control, a Reflection

My Experience

I remember growing up playing racing games on my phone, steering using the phone was my first real experience ‘operating’ the controls of a virtual vehicle. Before I ever stepped into the driver seat of a real automobile, I was driving virtual vehicles using my phone, joysticks, and replica steering wheel and pedals in games like need for speed, nascar, forza, ect. To me and my generation, this was a normal experience that we tend not to think back on as being such an amazing leap, technology has greatly impacted the way our generation and future generations experience controlling vehicles for the first time. In video games however, the vehicle dynamics experienced are far different than when operating a motor vehicle. Negotiating intersections, on ramps, merging onto highways, and avoiding sudden obstacles in the roadway are all real situations where there is potential for overshooting and undershooting the correct sequence of throttle, brake, and steering to achieve a safe desired final pose (position and orientation). Video games and simulations cannot entirely capture the true road surface conditions, nor can it perfectly replicate the behavior of other drivers on the road. Using handheld electronic controls that provide no feedback from the physical system can provide unrealistic expectations in terms of what to expect from the vehicle in response to a rapid rate of change in yaw for instance. The frequency at which demanded control values are sent to the vehicle in a simulation or video game rarely account for the physical limitations of the system, where in cases of overshooting, slip would occur and loss of control would ensue. Simply put, if we desire to change the way we control vehicles then we must change the way we prepare drivers in operating them in a realistic environment. I think that this paper brings a valuable argument to the table, which is that disabled drivers would have greater accessibility to vehicles if they were able to semi-autonomously operate them via a familiar control device such as a cellphone [1]. There is no harm in finding ways to adapt current technologies to existing problems in order to overcome an obstacle. However, there needs to be rigorous validation and testing accompanying the proposed solution.

Purpose and Expectations

Prior to reading this paper, I thought that the purpose would be to present evidence of a solution that would control vehicles using haptic devices. The haptic device in this case was in fact a cellphone, and the reason this device was chosen was due to its widespread availability and commonality amongst manufacturers in using accelerometers within these mobile devices. I expected perhaps greater discussion surrounding the implementation of this on a mobile device, but the papers aim was more targeted towards studying the general control and overall performance of the haptic vehicle control system. I was not entirely sure what to expect from the testing and results sections prior to reading the paper, since there was much to be determined without at least reading the abstract, the title leaves room for imagination.

Reflections

The methods section does well in providing clear derivations of the relevant dynamics and how the haptic controls are fed into the dynamic model. I thought that it was all very well thought out, there was a clearly presented hypothesis prior to giving details surrounding the experiment itself. This prepared readers mentally in getting them thinking about why the results correlated with the overall problem. I found myself very interested in the paper, it was easy to understand and there was little to no error in the analytical methods used to determine the results. There was sound justification at every presented conclusion, such that it was clear the assessment contained little to no bias. There was however, little discussion surrounding the issues associated with a haptic device being used and how the dynamics of the vehicle on the road may play into the failure of such a solution. For example, imagine you hit a pothole and all of a sudden you are demanding a sudden change in steering angle. I hope its not so much that you lose control? Or for instance, you lose grip of the phone and drop it on the floor. Good luck picking it back up while your car is responding to the sudden dramatic changes in control of steering, throttle or braking. It was clearly stated at the end of the paper that there is immeasurably much more testing and justification that must take place prior to this solution being implemented in OEM vehicles. I just think that there may be some more fundamental issues that may have been overlooked from the onset of the proposal that should have been addressed early on in the design of the experiment to set these concerns to rest.

Next Steps Moving Forward

The investigation presented proposed the use of the haptic controller in a semi-autonomous vehicle, however with the recent deployment of autonomous vehicles it seems more likely that this device would be unnecessary. There could be an adaptation of the research to push towards an investigation of what levels of comfort humans have when being driven by an autonomous vehicle. Perhaps by recording the inputs of drivers using a higher bandwidth haptic controller, we can gain insight into the kinds of constraints that should be programmed into the controllers which perform the path planning aspects of autonomous driving. The range of obstacle avoidance testing performed for an autonomous system should be far rigorous than the moose test performed in this experiment. The moose test did provide a good preliminary benchmark for this paper.

Conclusion

In summary, haptic devices are great candidates to solve the proposed problem of making driving vehicles more accessible for drivers with disabilities since it is lower effort and has shown to have promise in the level of error observed in preliminary experiments. I agree with the authors that more research needs to be conducted before there are actual implementations of this technology in real vehicles. Further research into solutions that can provide users with a way to override autonomous vehicles without the use of a steering wheel would open up more opportunities for a safer cabin space in terms of crashworthiness, however the proposed solution should address the aforementioned flaws in the current design.  

References

[1] Wang, C., Alexander, K., Pidgeon, P., and Wagner, J., “Use of Cellphones as Alternative Driver Inputs in Passenger Vehicles,” SAE Technical Paper 2019-01-1239, 2019, doi:10.4271/2019-01-1239.