Whitepaper on: The current status of Self-Driving Trucks and its challenges

Corresponding author: Dlshad Othman (e-mail: [email protected])

Date of publication: 08/07/2019, written on 02/01/2018

Background

According to the American Trucking Associations, the shortage of qualified truck drivers in 2019 reached the level of 59,500 drivers and it will increase to 160,000 by 2028[1].

This shortage created a good opportunity for self-driving car makers, to navigate the market and seek the opportunity to manufacture self-driving trucks, that can fill the gap and introduce new products to the market, that will allow moving trucks on the highways with limited interactions from drivers or without a driver, by building full automated self-driving trucks (Halsey 2017). Self-driving trucks are already being tested in the U.S., in 2016, Otto’s first self-driving truck successfully delivered a shipment of beer on a 120-mile trip, while the driver is setting in the backseat and observing the behaviour of the automated truck (Davies 2016).

The self-driving vehicles industry in general is on the rise, with 44 companies working on building their own autonomous cars (Cbinsights 2017), meanwhile companies like Daimler, Otto, Tesla and Volvo are working on producing new self-driving trucks with different levels of automations that range between assisting the driver to increase their safety, to a full autonomous truck (Muoio 2017). The American Society of Automotive Engineers identified six different levels of vehicle automation, starting with zero which has no automation to five with full automation (SAE International 2016). Nowadays, the market offers trucks at level one with limited driver assistance technology, and many research and development efforts are happening to build and introduce new technologies, hoping that will produce higher level of automated trucks, introducing new market and regulation challenges.There is no big difference between the required technology to build an autonomous car or a truck, they are both relying on advanced machine learning, sensors and communication systems as the base of a complex system that will make the care make decisions on the road. On the other hand, differences will start to emerge between both products when assessing the potential markets and trying to understand the legislative frameworks for both the business and transportation laws and regulations (Halsey 2017).

For the past seven years, 29 U.S states passed their own laws and regulations that govern self-driving vehicles. In 2017, the U.S Congress passed the (SELF DRIVE Act), to cover the self-driving cars regulations on the federal level. However, due to the complexity of the issue, and the pushback from the unions, the act didn’t regulate the self-driving trucks, and only passed regulations related to personal cars (“Self-Driving Vehicles Enacted Legislation” 2019).

The Truck industry in the U.S sees self-driving trucks as a threat to the 3.5 million truck drivers (Hassler 2017) which already mobilized drivers and unions to advocate against autonomous trucks. On the other hand, a more possibly damaging cyber threat is growing, in parallel with the rise of computer controls and digital communications over cars, and several proofs of concepts took place in the last couple years (Tuttle 2017). beside all of that, there will be a need to improve the infrastructure of roads and highways before allowing trucks on the road. The market of self-driving trucks is promising. Companies like Uber, Amazon, USPS, UPS and PepsiCo are funding R&D and per-ordering trucks from manufacturers that announced plans to produce self-driving trucks.

Technology roadmap

To create a roadmap that captures the status of self-driving trucks and maps the possibilities and challenges of its future, flexibility is needed. Hence, roadmaps are flexible when it comes to their format, purpose and use (Phaal et al 2004). This flexibility will allow adopting and somehow customizing the roadmap to reflect the best visualization of self-driving trucks evaluation process.

The first layer of the map is the Level of Automation which represent milestones and eras of self-driving trucks, in other words “levels of vehicle automation” according to SAE International. Bars were used to present the level of automation, and to show the available technologies at a specific time. The end goal of this map is the actual product which is placed on the second layer of the roadmap, to understand the technology evaluation, and market and legislative are at the lower layers.

Figure. 1. Self-driving Trucks RoadMap

The mapping process for Self-driving trucks has different current and future product characteristics, ranging from partly automated to fully automated products. Capturing the evaluation of the involved technologies and the challenges are represented in multiple layers or format (a) (Phaal et al 2004) that can be expanded and has sublayers roadmap.

Figure. 1. Shows a multiple-layers roadmap for self-driving trucks that was designed after an extensive research on self-driving history, current status and its future. This can be seen in the ‘Level of Automation’ or milestones layer that shows different eras of self-driving, starting with the current level that products are being manufactured at ‘Driver Assistance’ which by itself represent a different level of truck autonomous that. For example, Ford F150 truck, was the first truck introduced (Truck Adaptive Cruise Control System) which represent the least advanced technology in driver assistance level compared to Volvo Refuse Truck which can move the steering wheel and make some decisions, as the most advanced in the driver assistance level.

Although the technology is advanced today, and several tests took place on the streets for auto-driving trucks – which can be seen on the roadmap – other challenges are slowing the technology readiness and prevent the process from moving forward to an advanced era of automation. These challenges can be found in at least one out of four domains as Heslop et al. (2001) suggested, which can shape the strength of the technology. For example, the concept of self-driving in general is a new one, and it’s mostly relying on machine cumulative learning, which is still elementary and limited to only handle easy driving tasks, which puts technology builders in front of the learning challenge, which takes time and effort. On the other hand, sensors are still not advanced enough to function during bad weather. 

The market of self-driving trucks is the most important element in the process, which is not the case in self-driving cars as the decision at the end will be personal compared to self-driving trucks that needs a market approval, adoption and flexibility from trucks manufacturers and future clients. This can be done by capturing how the market is engaging with the topic to comprehend the size of investment that will be put in R&D at this stage and the production in the future. The map captures the ‘The Market Attractiveness’ (Heslop et al. 2001) including current and future deals and how such deals and demand are pushing the development of self-driving trucks.

Governments have power over businesses and manufacturers and they regulate transportation which puts the legal/legislative aspect of self-driving trucks in a very essential position. Especially, that there are already many loud voices against regulating self-driving trucks, mainly fuelled by fear of drivers losing their jobs and cyber security threats.

How the government is dealing with the issue and how new government bodies are being formed for this purpose, can impact the technology and future products. Therefore, placing them on the roadmap can show the direct impact on the final product and its future development.

Key challenges and recommendation for the technology

  • Advocacy, Public acceptance for this technology is needed, and this can be achieved by answering people’s related fears to self-driving trucks. This can be done by protecting drivers’ jobs at least for the next 10 years by adopting regulations that prevents any truck from moving without a driver, even if the driver is not actually driving the car. Such action can ease the pressure on the technology makers, so they can focus on achieving their goals. 

Also, increase the amount of safety research that can clear the fear of auto-driving machines in the public eye, by proving its capabilities of being safer than human drivers. These issues can be found on the map, ongoing safety research and no trucks without driver on the street regulation.

  • Legal Protection, Currently, the U.S government is avoiding regulating Self-Driving trucks under the pressure of Truck Drivers Unions and other lobbying groups opposing the technology, this was obvious in the SELF DRIVE Act 2017, which didn’t include any self-driving trucks regulations. But at the same time, the government is not preventing developing such technology or even testing it on the roads in most states.

For Self-Driving Trucks industry to move forward, a ‘Legal protection for autonomous trucking’ is needed to allow the market to move forward. At the same time, more research, tests and transparency are needed from the manufacturers to allow making better decisions on both the government and public level.

  • More data is needed, Machine learning systems need more data to offer quality results. Currently, labs rely mostly on simulations to generate data that can improve the AI part of the self-driving truck (Stewart 2017). However, to collect an out-of-lab data, an addons technology is needed to be installed on regular trucks to watch drivers’ behaviour and collect information. At the same time, this will give drivers the opportunity to be part of the evaluation process, instead of being excluded from process which they fear the most. This technology is reflected on the roadmap at the technology level in both knowledge information system and drivers improving machine learning.
  • Cyber Security is one of the mandates of the new U.S auto-driving legislation, under “Cybersecurity of Automated Driving Systems” which prevents manufactures of selling any auto-driving car without Cyber Security policy and a system that is capable of identifying, assessing, and mitigating any cyber-attack and prevent or correct in case an attempt took place (SELF DRIVE Act 2017). Same rules are going to be applied on trucks in case trucks were added to the Self-Drive act in the future. It’s important for the future of the technology to be ahead of the game by establishing new business model of threat intelligence that focus on self-driving cars, and studies attack techniques, malicious codes, vulnerabilities and auditing processes. The roadmap represents this issue in two different points, Threat Intel Cybersecurity Centre and The Era of Cybersecurity Technology in Design which is getting cybersecurity experts to be more involved in the designing stage, to build secure technology rather than investing on securing it afterwards.
  • Improved Vehicle-to-vehicle communication system, In December 2016, the Obama administration proposed a mandate to add Vehicle to Vehicle Technology (V2V) to new cars which will allow cars to talk to each other wirelessly and share instant data aiming to prevent traffic accidents (NHTSA 2016). Such technology can be essential to self-driving cars/trucks. However, the proposed mandate was quietly put aside by Trump administration recently (Lowy 2017). V2V is an important part of trucks automation process, and by having government making it a mandate is a good opportunity to increase V2V within larger number of cars, and this will make roads safer and more information will be flying over the highways. The issue can be found on the technology level of the roadmap, Improved V2V system, also by mandating V2V system by the government.
  • Autopilot is when a truck is fully capable of driving under a fully autonomous mode, with or without supervision from a driver. In this stage, technology will be very advanced to the point that minimum interaction from humans is needed. At the same time, humans can start building trust in the machine by watching and using it. The roadmap represents this issue thru the following points, capturing more Data, that will improve the AI level and increase the safety and reliability of the trucks. New insurance Market will be established to match expectations of future clients and the market changes. Also, the government will be Introducing Autonomous Vehicles Lanes that will restricted self-driving vehicles to specific lanes, to assure regular drivers their safety and give them at the same time the opportunity to witness how automated cars are driving on the roads. This will also be beneficial to stakeholders by giving them more accurate and tested data on automated cars’ behaviour.
  • Self-driving trucks are superior to average human drivers, this can be the goal of this technology, and if self-driving industry was able to achieve it then self-driving trucks will become fully autonomous. In their ‘Autonomous Vehicle Technology Guide for Policymakers’, Rand Corporation, recommended that full autonomous cars can be permitted only if they superior to average human drivers (Anderson et al 2016). This was presented on the map as a level of automation or milestone that can announce an era of fully autonomous trucks. 
  •  

Conclusion

Solving the technology challenges of Self-Driving trucks might not be a hard task compared to the other legal and advocacy challenges. Governments (Federal and state) need to understand the technology and its regulatory aspects to produce the needed legal protection. On the other hand, a deal is needed with Truck drivers’ unions and organizations, in order to include them in the process of building this technology, as there are tools that can guarantee their jobs for the near future, and that the main goal of this technology is actually to close the gap between the market and the demand, which is not going away anytime soon

References

Halsey, A. (2017) As the era of driverless cars looms, can self-driving trucks be far behind? – The Washington Post [Online]. Available at http://wapo.st/2eWtOLN?tid=ss_tw-bottom&utm_term=.0701ca21face  (Accessed 3 January 2018).

Davies , Alex (2016) Uber’s Self-Driving Truck Startup Otto Makes Its First Delivery | WIRED [Online]. Available at https://www.wired.com/2016/10/ubers-self-driving-truck-makes-first-delivery-50000-beers/ (Accessed 8 January 2018).

Cbinsights (2017) 44 Corporations Working on Autonomous Vehicles [Online]. Available at https://www.cbinsights.com/research/autonomous-driverless-vehicles-corporations-list/ (Accessed 8 January 2018).

Muoio, Danielle (2017) Autonomous trucks by Tesla, Uber, Google will change trucking industry – Business Insider [Online]. Available at http://www.businessinsider.com/autonomous-trucks-tesla-uber-google-2017-6/#uber-is-pursuing-self-driving-trucks-through-otto-a-startup-the-company-acquired-last-august-but-the-project-is-at-the-center-of-a-massive-lawsuit-filed-by-waymo-googles-sister-com (Accessed 8 January 2018).

SAE International (2016) Taxonomy and definitions for terms related to driving automation systems for on-road motor vehicles J3016. http://standards.sae.org/j3016_201609/ (Accessed 8 January 2018).

NCSL (2018) Autonomous Vehicles | Self-Driving Vehicles Enacted Legislation [Online]. Available at http://www.ncsl.org/research/transportation/autonomous-vehicles-self-driving-vehicles-enacted-legislation.aspx (Accessed 8 January 2018).

Hassler, S. (2017) ‘Self-driving cars and trucks are on the move [Spectral Lines]’, IEEE Spectrum, vol. 54, no. 1, pp. 6–6 [Online]. DOI: 10.1109/MSPEC.2017.7802341. Available at http://ieeexplore.ieee.org/document/7802341/ (Accessed 8 January 2018).

Tuttle, H. (2017) ‘Risk Management – Hacking Cars’, Risk Management, vol. Vol.64, no. (1), p. p.20(6) [Online]. Available at http://www.rmmagazine.com/2017/02/01/hacking-cars/ (Accessed 8 January 2018).

Phaal, R., Farrukh, C.J.P. and Probert, D.R. (2004) ‘Technology roadmapping – a planning framework for evolution and revolution’, Technological Forecasting and Social Change vol. 71, nos 1–2, pp. 5–26. View the paper at http://www.open.ac.uk/libraryservices/resource/doi:10.1016/s0040-1625(03)00072-6  (Accessed 8 January 2018).

Jack STEWART (2017) Why Daimler Researchers Used VR to Become Self-Driving Cars | WIRED [Online]. Available at https://www.wired.com/story/moovel-self-driving-car-experiment/ (Accessed 10 January 2018).

LOWY, J (2017) APNews ‘Break: Gov’t won’t pursue talking car mandate’ [Online]. Available at https://apnews.com/9a605019eeba4ad2934741091105de42  (Accessed 10 January 2018).

Anderson, J., Kalra, N., Stanley, K., Sorensen, P., Samaras, C. and Oluwatola, O. (2016) Autonomous Vehicle Technology: A Guide for Policymakers, RAND Corporation [Online]. DOI: 10.7249/RR443-2 (Accessed 10 January 2018).


[1] Journal, C. (2019). ATA trumpets persistence of shortage of qualified drivers. [online] Commercial Carrier Journal. Available at:

ATA trumpets persistence of shortage of qualified drivers
[Accessed 7 Aug. 2019].

Share Button

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.