By Kevin Armstrong
Elon Musk has already promised additional improvements to Tesla’s full self-driving beta shortly after AI Day on September 30. Now we get some details on what 10.69.3 will include.
The FSD Beta group grew by 60% in the United States and Canada when Tesla reduced the safety score requirement from 90 to 80. There are now about 160,000 Tesla owners in the program. The jump means more enthusiasts are giving FSD comments on Musk’s favorite platform – Twitter.
Chuck Cook, the man behind Chuck’s infamous corner, also known as the unprotected left turn, took the latest FSD for a ride to its most tested spot. His YouTube videos are fantastic, especially the unprotected left-turn videos that show multiple camera angles of his Tesla navigating that tricky spot. He posted a video on Twitter showing 10.69.2.2 successfully piloting through the area, which caused Cook to cheer and cheer, “wow, yeah!”
Improvements in small gaps
Unsurprisingly, Musk saw the video. He pointed to Cook at the recent shareholder meeting and said he welcomed the fair and honest criticism of the turn and said it would be dealt with. The CEO replied to the video: “The car will move on tighter gaps as we improve NN (neural networks) speed predictions for crossing traffic. 10.69.3 next month has some incremental improvements.” NN stands for neural networks, and this response shows that despite Cook’s endorsement, Musk is confident the system can and will do better.
Turn signal improvements
Twitter user @Johnkrausephotos tweeted: “I love FSD Beta – but – could the vehicle trigger the turn signal before entering the turn lane? It uses the signal for speed dependent lane changes, but not merging into a lane of turn.” The tweet caused a lot of reactions as some users criticized the company for programming its cars to change lanes before reporting. Musk must have been on the same page as he responded with a simple “Yes” and no further explanation.
Match speed to traffic
Silicon Valley Tesla Owners took to Twitter to ask“Elon, when will the tesla speed up when it notices traffic speeding up? Sometimes it slows down.” Again, Musk had a quick and short response, tweeting, “Next month.” We can only guess that this will also be part of 10.69.3.
This trio of improvements should be uploaded to Teslas with FSD right after AI Day. So, as the world awaits the unveiling of Optimus, the humanoid robot, Tesla owners will be eager to see what 10.69.3 will bring to the vehicle.
By Kevin Armstrong
Tesla owners can say goodbye to range anxiety. Despite Tesla’s extensive network of more than 35,000 boost stations, there are still areas that can increase stress levels as the load decreases.
Those days will soon be a distant memory as new non-Tesla charging stations appear. However, these stations don’t have the sleek Tesla charger, but a big, bulky one called CCS.
CCS, which stands for Combine Charging System, comes in two types. Type 1 is mainly used in North America and South Korea, while CCS type 2 is the charging standard in Europe. They differ in the number of phases and the maximum power they can provide.
Until now, Tesla did not sell its Type 1 CCS adapter in North America, but in South Korea. If you wanted one, you had to import it or buy the adapter through a third party. These adapters fetch a high price online, including on eBay. Some sellers sell the adapter for up to $1,000.
However, Tesla eventually added the adapter to its US store. Tesla owners will now be able to access an entirely different charging network.
Tesla showed off the device in the store for the bargain price of US$250, a significantly better deal.
But not all Teslas are CCS-capable. Some models can charge with CCS once the adapter is connected. However, several models, especially versions prior to 2020, are not CCS ready. There is an easy way to find out if your Tesla is CCS compatible by tapping on the main menu, then Software, then Additional Vehicle Information. A box will appear with lots of information; look for CCS adapter support. It will show ‘Not Installed’ or ‘CCS Enabled’.
Electronic control units (ECUs) were available in the Tesla parts catalog so owners could retrofit their cars if they were incompatible. However, after the adapter launches in the Tesla store, visitors are prompted to log in before they can purchase the adapter.
The website determined if the owner had a CCS-ready Tesla. If you have not done so, the following message is displayed: “This accessory requires an upgrade. Check back early 2023 for availability. Further down the page, there are instructions for checking in early 2023 for availability of vehicles requiring an upgrade.
Tesla has also added messaging to its mobile app. If you navigate to the Service section, you will now see a message at the top of the application displaying “CCS Adapter Retrofit, check vehicle status”.
Unfortunately, tapping on it offers no additional information, asking owners to check back in early 2023 to schedule an appointment for a renovation.
CCS stations are reported to reach speeds of over 150kW in the United States and, unfortunately, they are the preferred outlet of many manufacturers. There was a petition to make Tesla charging hardware the industry standard, but it seems to have failed. CCS is available from Electrify America and Electrify Canada, and several other third-party electricity providers.
By Kevin Armstrong
Tesla may be stepping up how it uses simulation to train its Autopilot system. A report from Electrek claims it has sources saying the company is focused on a San Francisco replica. The article includes an image of the recreation and indicates that Tesla is working with Real Engine on its simulation.
According to Electrek, the image below is part of Tesla’s simulation of San Francisco.
Tesla gave the world a glimpse of how it’s using simulation to drive the Autopilot program forward on AI Day One in August 2021 (recap).
At the first AI Day, Tesla talked about using simulations to help train Autopilot. The video below shows where they discuss a simulation.
Ashok Elluswamy, the Autopilot Program Manager, showed a video that at first glance appeared to be real other than an appearance of a Cybertruck. “I can say that myself. It’s very pretty,” Elluswamy said. He explained that the company is investing heavily in the use of simulation. “It helps when the data is hard to find. As big as our fleet is (FSD Beta users) it can still be difficult to get crazy scenes,” the director explained, showing a render of two people and a dog running in the middle of a busy highway. . “It’s a rare scene, but it can happen, and the autopilot should always handle it when it happens,” Elluswamy said.
It looks like Tesla has jumped on the Fortnite Battle Bus by partnering up with Epic Games and its development platform – Unreal Engine. Fortnite is one of the most popular games of all time, with 80 million subscribers and 4 million daily users, and it was made with Unreal Engine. Epic flexed its creative muscles when it brought together experts to create The Matrix Awakens: An Unreal Engine 5 Experience. The aim was to “blur the lines between cinema and play, inviting us to ask ourselves – what is real?” The project’s spotlight on Unreal Engine shows just how incredibly realistic a simulation can be.
After Elluswamy explained that the company invests in simulation, it makes sense that Tesla would hire several positions with simulation in the job description. Electrek highlighted an ad for the Autopilot Render Engineer. The publication states that the successful candidate “will contribute to the development of autopilot simulation by enabling and supporting the creation of photorealistic 3D scenes capable of accurately modeling the driving experience in a wide range of locations and conditions.” Tesla prefers applicants to have experience working with Unreal Engine.
While this isn’t new, it shows that Tesla is stepping up its efforts to improve Autopilot. It recently rolled out Full Self Driving to an additional 60,000 users, bringing the FSD Beta program to 160,000 in North America.
We can only guess how many thousands of simulations the Autopilot team runs to add to the data that beta testers collect. It seems unlikely that Tesla only created the City by the Bay in his simulations. Elluswamy may show more renders on Day 2 of AI on September 30.