Scaling To Meet Engineering Challenges In Transportation

If you’re working on anything related to self-driving cars, you’re likely pondering the tradeoff between what’s technically possible and socially feasible. Example: what do you do if the posted speed is 25mph while the local norms are to travel at least 30 mph? Obey the law and annoy the drivers around you? Or follow the herd and risk the ire of local law enforcement and officials who are key stakeholders in the ongoing testing and rollout of driverless solutions?

Expect this tension to be a fact of life over the next many years if not more. The reason for hedging on timeframe is that after years of optimistic predictions about the approaching zero hour when Autonomous Level 4/5 vehicles will arrive, it’s become clear that human-driven vehicles will be the dominant modality for the foreseeable future. Note that I didn’t say human-driven cars. Micromobility, including e-bikes and scooters, has exploded onto the scene, capturing media attention, VC funding and a scurry of acquisitions by established OEMs. Indeed, micromobility today is reliably described as a “revolution,” much as self-driving cars were a few years back.

And transit authorities and cities loom especially large as well, given the macro trend of urbanization around the globe. Officials like Seleta Reynolds, general manager of L.A. Department of Transportation, are emerging as prominent foils to the tech industry. See for example this thread spawned by Reynolds poking at some of the Boring Company’s claims in vernacular that tech faithful and nerds can appreciate.

The cost of the tunnel is low because the tunnel is small. You need dozens of tunnels for non-stop A to B trips (and to solve for the “red herring” of induced demand what). At what point does the tunnel of tunnels reach the mines of Moria and the cost point of an actual subway?

— Seleta Reynolds (@seletajewel) December 19, 2018

Among the constants in the ongoing tech upheaval of transportation is the importance of sensors, critical in everything from by-now relatively basic ADAS features like basic adaptive cruise control (ACC), automatic parking and blind spot monitoring, to the more robust prototype autonomous vehicles with full-stack technology, including multiple sensor modalities and robust sensor fusion and path planning capabilities, powered by advanced machine learning and AI. Of course AI gets all the attention, but many expect renewed attention on ADAS given some of autonomy’s setbacks last year.

Here’s VSI Labs founder, Phil Magney, quoted in an article last month on the return of ADAS: “The rollout of AV technologies is a lot harder than people realized compared to the lofty targets first established. The automotive industry has a renewed interest in ADAS. Call it ADAS 2.0 if you like.”

Micromobility depends on an expanding array of sensors, as well. Today it’s the availability of cheap torque and various other motor-related sensors, GPS and gyros that have supercharged that segment. In the future, onboard cameras might help keep dockless scooters and e-bikes safe from vandalism.

One reason this current mobility moment is so exciting is that, broadly speaking, the industry is transitioning from mostly off-the-shelf sensing hardware and software to bespoke solutions specifically tailored to self-driving and other applications. An example is the astonishing advances in imaging chips to meet requirements for spatial and angular resolution, dynamic range and bit depth that are central to seeing all the car in all conditions.

Another example is work being done to update the AUTOSAR methodology for autonomy, a topic discussed in our recent podcast, this time featuring Mathias Fritzson in Gothenburg, Sweden. And look for a blog post shortly on a major automotive OEM approving our Volcano VSTAR product for designing its next-generation ECUs.

Within Siemens PLM, we are excited to be enabling this transition from science project to deployment. The lesson in mobility so far is that, unlike other tech realms (especially in software and app development) where a small handful or even one engineer can achieve outsized results, creating new mobility solutions is a team effort cutting across domains and requiring scale. Computer science, robotics, hardware and software development, cloud computing, AI and machine learning, safety-related testing and maintenance, urban planning and traffic management — the list of relevant disciplines goes on and on.

This week CES is dominating the tech headlines, especially when it comes to automotive and transportation. The meta-story is chip-to-city. I’d wager nearly all the transportation tech vendors showing off their wares fall somewhere on this spectrum, though very few have offerings that solve problems spanning submicron IC design to urban/suburban commuters. I’ll have more to say on thinking about the diverse yet highly interrelated scale of the engineering challenges in transportation in the coming months.

Happy 2019. And if you can, tweet me with biggest chip-to-city story you’ll be watching this year @AndyMacleod_MG.