Andrew Jones

After some exchanges in the post below, Antonello De Galizia, Ph.D. and I had an interesting talk on safe #automateddriving. Because the community indeed struggles to navigate the complexity of this domain and to call things by their name, we didn’t want to spread the confusion and leave the discussion unsolved. We found out to be very well aligned and thought it was worth sharing our conclusions with you:   🚗 While both ISO26262 and ISO21448 are strong guidelines for ensuring safety during the development and design of automated driving system, they address different sources of risk, and these should be all tacked according to best practices and state of the art engineering. 🚗 While both standards are in place and provide the key concepts for safe automated driving, the complexity is in their implementation. This is especially true in the case of ISO21448, due to the lack hands-on experience. For example, investigating all possible scenarios within the target operational domain (TOD), considering all relevant combinations of the relevant factors (e.g., environmental conditions) is key to study the system functionalities and claim absence of unreasonable risk. 🚗 Following (to the letter) these standards is not sufficient to achieve truly safety automated driving. In fact, limiting the safety process only to cases where a malfunctioning (26262) or hazardous behavior (21448) result in safety relevant hazards (i.e., severity classes S1, S2 and S3) would mean that situations where unpredictable, irresponsible, or unlawful driving actions would be neglected if not found to be directly leading to harm. These need to be addressed too for a much more solid and proactive safety approach.   We hope to have contributed to a little more structured discussion on safety in this way. Feel free to reach out, Antonello & Henk Here is the link to the original post including our discussion: https://lnkd.in/e8ekRTtp

23

What can we do that’s fundamentally different with AI, cloud, and compute technologies to transform how people get around? That’s why I joined General Motors a year ago, and it's what drives our team to take on some of today’s biggest engineering challenges. Motor Trend recently visited our tech center to talk about how we’re making higher quality, longer-lasting, easily-upgraded vehicles for their Coding the Car series. Check it out: https://lnkd.in/gpk2b76V #SDV

138

5 Comments

Second day summary of the Connected Vehicle Systems Alliance (COVESA) AMM meeting: Started the day strong with an overview on how Amazon has succesfully integrated #Alexa in vehicles. Arianne Walker presented Amazon’s roadmap and achievements with using Voice commands in vehicles. Switching tracks - it was interesting to see Hisao Munakata’s view on how to migrate RTOS SW to an HPC- important topic for SDV development. The COVESA member visibility round was time well spent - discussing with Carin Lagerstedt and Brian Carlson ways to enable COVESA members be more visible within the organisation - we’re lookimg forward to get our marketing team from SMART Engineering & Development working on this topic. The afternoon was reserved to exploring how AI can be integrated within vehicle. Great panel with Jeff Blackburn, Tim Yerdon, Richard Fernandes, Jean-Luis Rivard and Ephrem Anthony Chemaly PhD. The secod day closed with a fireside chat (minus the fire) between Keith O’Neil from Globant and Brian Wassom. This concluded day two and this AMM. Looking forward to next year’s event. I am returning home hopefulabout our future but also aware of the ammount of work and effort we need to put in to change this industry towards better sustainability. #COVESA #automotive #AI #sustainability #softwaredevelopment

16

4 Comments

Examining Claims About Waymo's Autonomous Vehicles Several people are raising concerns about the development and deployment of Waymo's autonomous vehicles, including accusations of inadequate transparency, use of guinea pigs, and inferior simulation technology. These are serious claims meriting thoughtful analysis, especially for a technology with such potential impact on public safety. I admit that autonomous vehicle technology is not my specific area of expertise. As an excited but earnest observer, I clearly lack key perspectives or insights someone working directly in the field would have. Rather than dismiss worrisome warnings, I aim to cut through accusations to find truthful feedback on how to enable innovation while protecting people. Based on current third-party research, Waymo seems to demonstrate safety levels exceeding human driving capabilities: - An extensive 2021 study by Swiss Re, a highly reputable global reinsurer, analyzed over 3.8 million miles of rider-only autonomous driving by Waymo and found zero bodily injury claims, along with reduced property damage claim frequency. - Additional findings from Waymo compare their vehicles' safety record over millions of miles to real-world human driving benchmarks. These indicate their system effectively avoids collisions and resulting injuries at rates far surpassing human drivers. - There are no clear indications that Waymo uses subpar simulation practices. Their published details highlight advanced proprietary simulation capabilities for modeling sensor data and scenarios. While not definitive, this research implies Waymo prioritizes safety and due diligence during development. Transparency and accountability remain crucial as the technology evolves. The posters have not cited any specific data sources that expose deficiencies in Waymo's approach, however the background of those commenting suggest they are experts on the topic so I would like to get to the bottom of it without fear mongering. When discussing these items, I’m always looking for evidence-based discussions on these issues. If there are flaws in Waymo's methodologies or gaps in their public reporting, sharing these is highly productive. As this technology matures, our discourse and decision-making processes should reflect the same rigor and care that we expect from the engineers building autonomous systems. - MIT Technology Review: [Article on Waymo's Virtual Testing](https://lnkd.in/gKnXrvVp) , Will Knight has brought important insider's viewpoints to the conversation around transparency in AV development - Gadgets 360: [Waymo Trains and Tests Its Autonomous Vehicles](https://lnkd.in/g9GnFf44) - Autonomous Vehicle International: [Waymo's SimulationCity AD Evaluation Tool](https://lnkd.in/gMmKr8-q) #waymo

8

9 Comments

Curious about how the Puget Mobile 17" stacks up against the M3 Max MacBook Pro in terms of battery life and thermals? Here's what we found: 🔹 Battery Life: MacBook Pro excels in light workloads, lasting all day, whereas the Puget Mobile 17" is better for half a day of light tasks. For heavy workloads, both laptops last about an hour. 🔹 Thermal Performance: Under light tasks, the Puget Mobile is cooler. For heavy tasks, the MacBook Pro remains comfortable for lap use, while the Puget Mobile can get hot. 🔹 System Noise: The MacBook Pro is quieter under heavy load (54 dBA) compared to the Puget Mobile (60 dBA). Find out more in our detailed analysis! ➡️ https://hubs.ly/Q02K3fmx0

3

Is the complexity of EVs and AVs overwhelming your team? Discover how model-based systems engineering and generative design can revolutionize your vehicle architecture. Explore how MBSE and generative design can: Streamline the process of integrating new technologies Reduce development cycles and get your products to market faster Identify the most efficient and effective designs Don't miss out on this opportunity to stay ahead of the curve. Read our blog post to learn more: https://sie.ag/6aHnr1

9

Waymo driving on the wrong side of the road due to what appears to be a bad driving decision -- for the second video in a week. Wonder what their excuse will be this time! Last time it was a strained story that they were trying to be safe by moving into an oncoming traffic lane (with oncoming traffic) for well over a block a few days ago (pointer to that earlier video here; involves a swarm of unicycles: https://lnkd.in/eHFicWNg ) This video is a lot more straightforward. The problem with heavy-tail situations is it highlights a lack of "common sense" that is an inherent limitation to machine learning-based technology. Not a great look for a company trying to convince us that human drivers are terrible. Video source is apparently from TikTok. Or a deepfake -- or one from a year ago that just surfaced -- let's see what Waymo says. Happy to update with more info on date/source/provenance if we find out. Update: I got a note saying this was in Phoenix from an informed source, but not iron-clad confirmation. The previous video with crazy unicycle circumstances, no crash, and a laughable explanation is one thing. A second video makes Waymo's one-off edge case defense a lot less credible. Source: https://lnkd.in/eQJQxsap

277

69 Comments

AI is driving the next wave of growth for the semiconductor industry (and those of us in the industry can't wait to see the anticipated growth 😉). Recently, I was encouraged to watch a keynote featuring Jensen Huang of NVIDIA. While it's highly informative and I recommend watching it in its entirety, what struck me most had nothing to do with AI. Around 36:05, the facilitator asked Jensen to provide advice on how to achieve success. His response is profound, drawing parallels between expectations, resilience, success, intelligence, and character. His commentary on how he leads his 55 direct reports (yes, 55!) is also thought-provoking. Take a few minutes to watch this keynote and listen to Jensen's insights on success and leadership. #artificialintelligence #NVIDIA #leadership

20

2 Comments

News about ADS - Jun 17, 2024: 😢 1) Waymo driverless vehicle contacts with a tele pole in an alley near 7th Avenue and Roosevelt Street 8 mph when pulling over in AZ on May 21, 2024 -> SW recall on Jun 12, 2024 https://lnkd.in/g9xQScah https://lnkd.in/gC_MYrE2 https://lnkd.in/g8zAxcQZ Root cause: "SW map" seems not a real root cause-> possible perception issue 2) Real robot driving a car - robot driver Misachi 😁 https://lnkd.in/ggpB_u3G 3) "NVIDIA Research Wins CVPR Autonomous Grand Challenge for End-to-End Driving" 👍 https://lnkd.in/ginE6CWF 4) Tier IV & Suzuki partner up for ADS 👍 https://lnkd.in/gTzvTrEJ 5) A sad story - but what happens to the driver who used the phone - too unresponsible 😔 - Tesla with FSD engaged from AP News https://lnkd.in/gCS9eXuZ

1

I'll go out on a limb and state... All performance measures for complex systems (like a product development team) are imperfect. That said, the following still holds true: 1️⃣ Just because a performance measure is imperfect doesn't mean it isn't useful and valuable. 2️⃣ Some performance measures are better for your context than others. Select the "best" measures for your context. Change them as you learn more about your context. 3️⃣ Using performance measure imperfection as a "reason" not to select them is irresponsible. Don't do this. #complexity #productdevelopment

5

2 Comments

This article gives a lot to think about. While software can achieve many things, it reminds us that creating a successful business case is not easy.

5

A study in perspectives on AV safety: a Waymo robotaxi approaches an intersection in San Francisco. Some other driver is doing "really fast donuts" in the intersection, and has been doing so long enough that the robotaxi rider looks up because of the smell of burning rubber entering the cabin. Robotaxi enters the intersection, then stops. The other driver halts and narrowly misses hitting the robotaxi (by perhaps a foot). The rider: "I can trust the Waymo, but I can't trust this driver". States that the other driver was upset but the robotaxi rider "shrugged, and pointed to the invisible driver." What ya' gonna do? It's a Waymo. "Still going to use Waymo." Presumed Waymo response to regulators and media: The human driver doing donuts was irresponsible. If there had been a crash it would have been the other driver's fault. (If I got this wrong would be happy to be corrected by Waymo. They have not commented as far as I know.) A broader safety engineering response: This is not the first time a robotaxi has encountered another car doing donuts in an intersection. In fact, a Cruise robotaxi was hit in a similar situation in March 2023 (SGO 30412-4960). From the description it seems there was plenty of warning this was happening. Why did the robotaxi not stop well clear of the intersection as a prudent human driver would have? Is Waymo monitoring SGO data and learning from what has been reported by other companies? Has Waymo included "other car doing donuts" in its hazard log? Did Waymo decide to ignore mitigation of this hazard for some reason? Will that change now that this has happened to Waymo instead of just Cruise? I'm glad in this case there was no injury, but there certainly could have been but for chance from a rare circumstance that is already known to have caused a crash. Being saved by the “grace of the Gods” is not a safety argument. I hope these questions are being asked and resolved by Waymo. But we won't know until the next donut incident. Primary Source: https://lnkd.in/eFHEMVFX Article: https://lnkd.in/eFVpfjec

65

14 Comments

Although I agree with the general principles behind the California "SB-1047 Safe and Secure Innovation for Frontier AI Models Act" there are some details that I disagree with. One of them is the commisioning of consultants in SEC.5 which adds section 11547.7 whose heading reads: "11547.7. (a) The Department of Technology shall commission consultants, pursuant to subdivision (b), to create a public cloud computing cluster, to be known as CalCompute, with the primary focus of conducting research into the safe and secure deployment of large-scale artificial intelligence models and fostering equitable innovation that includes, but is not limited to, all of the following: ... " The commisioning of consultants opens the door for a wide variety of conflicts of interest, some related to the high rates charged by consultants others related to hidden agendas and self interests. It is a much better idea that the "Department of Technology" has its own personnel with a high level of expertise in the field that would avoid this conflict of interest of industry consultants. By way of comparison, the Europeran Union AI act (EU AI Act) does not require consultants, instead Article 65 calls for the Establishment and Structure of the European Artificial Intelligence Board composed of one representative per Member State. BTW, some of the provisions of the EU AI Act went into effect Aug. 1, 2024 while the remaining ones will do gradually in the near future. In my opinion, the governance of the EU AI Act is better thought out than the CA 1047 bill. At the end of the day, an AI bill is probably better than no bill. What is your opinion?

4

Reposting to share historical context on previous work done by Carnegie Mellon related to AVs avoiding poles and trees. The research details are in the comments sections attached to the post, including cool videos from CMU NREC. The videos are dated, but clearly still relevant.

2

📈🚗Increasing levels of driving automation require system solutions that enable highly predictable, accurate and real-time response—and that’s #NXP's focus. Learn more about ADAS and Safe Driving on our web!

2

As rider-only mileage accumulates, so does the Safety Impact of the Waymo service. Proud to be contributing to traffic safety like this! Our safety research papers are the enabler for understanding the underlying methodologies and calculations. Find them at waymo.com/safety

51

1 Comment