Tesla V10.0 Update: More Entertainment Options and Improved Linux Compatibility

Tesla has finally released V10.0 system to North American early access program members after a year of cross-version updating. This update not only brings many new features but also sheds light on many more profound pieces of information.

Entertainment Section

Since the heavily anticipated Smart Summon feature is missing once again, let’s start with the entertainment section:

Tesla Theater

Tesla Theater provides access to iQiyi, Tencent Video (YouTube and Netflix overseas), and Tesla Tutorials, allowing for video playback on the car’s display while parked and connected to WiFi.

Ximalaya Audio

Chinese consumers have access to Ximalaya podcasts and other audio content through the media section in Tesla China.

Cuphead Tesla Edition

The classic shooting game is available, but requires the car to be parked and a USB gamepad to be connected to play.

Karaoke

Karaoke is also available and requires the car to be parked to be used.

Feeling Lucky or Hungry? (Delayed in China)

When you’re hungry or undecided about a destination, click Hungry or Lucky and let the Tesla map engine recommend a restaurant or tourist attraction, including reviews and descriptions.

The first impression of V10 is that Tesla is determined to push Linux content and ecosystem with its own power, to promote the prosperity of the Linux automotive content ecosystem.

Linux faces a natural disadvantage when it comes to products and content ecosystems in the entertainment industry. This is also reflected in V10: Due to the lack of Linux client versions for iQiyi and Tencent Video, Tesla had to implement them in the form of web versions on V10, since Tesla’s central OS is based on Linux. Earlier, QQ Music was introduced to Tesla’s central unit but could not implement all of its functions because of the Linux ecosystem’s lack of capabilities.The speed of Tesla’s localization in China is praiseworthy, with QQ Music, iQiyi, Tencent Video, Ximalaya… as mainstream applications and software that will gradually be used by more and more Chinese users on Tesla cars as mentioned by Tesla China’s President, Wang Hao.

In the V10 update, Tesla’s rushed work also reflects its localization efforts, such as the Chinese translation of the Tesla Tutorials. However, the name of the tutorial itself is still in English. Moreover, the “Feeling Lucky or Hungry?” feature is absent from the Chinese market. This feature seamless integrated with Dianping, a Chinese Yelp-like application; The reason for not launching this feature domestically might be due to challenges associated with transferring location-based services and cooperation with local companies.

If there is a difference between Tesla and leading domestic smart car manufacturers in the field of smart cockpit ecology, it is the belief in OTA over-the-air (OTA) upgrades.

Let me explain, for example, why smart car owners rarely watch videos or sing karaoke in their cars. It is simply because the user experience is not good. The screen resolution, audio quality, and lack of space makes it uncomfortable for users to use and unlikely to do it a second time.

Therefore, in the first step, car companies must equip their cars with high-quality hardware, including but not limited to CPU, GPU, screens, and audio systems, to lay a good foundation for information entertainment experience. Secondly, they should support OTA upgrades. Developers can provide better features and performance by maximizing the potential of these hardware specifications and performing complete and adaptable scenario and feature programming to ensure better playability.

Elon Musk’s previous criticism of the Model 3 audio system focused on its “highly programmable” capabilities, rather than audio quality, that allowed audio engineers to program OTA updates to improve performance constantly.

As expected, before the release of V10, Elon began to show off in public, “the combination of videos and karaoke experience with the Tesla sound system is amazing.”

To enhance playability, Tesla has also done some other things, such as landing games “Beach Buggy Racing 2” and “Cuphead” on the Tesla central control. In addition to the transplantation, Tesla has added some elements of Tesla or Space X to increase user freshness.

To sum up, although Linux does not have a prosperous content and ecology, Tesla has shown great tenacity in promoting the prosperity of the Linux smart car ecology by its own strength. From the perspective of implementation effects, Tesla has worked hard, but the effect is not very good. Not to mention the web video, the most basic map and payment functions are difficult to land on the Linux side.

On the other hand, Tesla’s cross-version updates have limited time and rough quality, but they can always be optimized to a good level. Maybe these features will become more and more fun and useful step by step before V11 is released.

Since choosing Linux, there is still a lot of work to be done from “usable” to “useful”. The game has just begun, and Tesla still needs to continue to work hard.

Assisted Driving

The end.

Well, this is a clever trick I’ve pulled. In the past three months, Elon’s evaluation of Smart Summon has changed from “the most challenging, good-looking, iconic, and magical in V10” to “almost perfect.” The delay of V10 version is completely attributed to Smart Summon, but at the last minute, Smart Summon was still removed.

You may want to say, if the automatic navigation driving assistance (NoA) in the high-speed scene can be iteratively optimized, how difficult can a semi-closed low-speed scene (typical application scenario for parking lot) of Smart Summon be?

Smart Summon refers to responding to the driver’s mobile app call from any parking space in the parking lot and driving automatically to the driver’s location. Although it is a semi-closed low-speed scene, in the process of functionality landing, powerful perception, decision-making and control capabilities are indispensable.

First is perception. Elon previously mentioned that there are two most challenging scenarios, one is the intersection with complex traffic lights, and the other is the intersection of a shopping center parking lot. With different vehicles entering/exiting, occasional pedestrians/staff passing by, narrow sections to perceive, unidentified objects to detect and recognize… Although the speed is slower, the perception challenge in a large parking lot is much higher than that in a closed high-speed scenario.

At the decision-making level, more challenges are exposed. For example, when entering/exiting the intersection, the system needs to understand the meaning of different vehicles’ turn signals and reverse lights, and even in many cases, the front camera needs to understand the meaning of gestures. The ability to interact with other vehicles is the real challenge.

At the control level, low speed requires more precise start/stop and speed control of the vehicle.

Intelligent Summon and Navigate on Autopilot (NoA) differ mainly in that there is no driver in the driver’s seat for the former, which brings huge challenges to the Tesla Autopilot team. This means that Tesla’s team’s agile, high-speed iteration mode has failed for the past three years.

Taking Navigate on Autopilot (NoA) as an example, Tesla initially released a version with a very poor experience. Regardless of entering/exiting the ramp or the automatic lane change logic and speed, they were all very poor. At this time, the driver would choose to take over, but at the same time, Tesla also collected a large number of brand new scenarios and optimized and fixed these bugs one by one in subsequent version updates, achieving the development of the function.

However, for Intelligent Summon, once there is unclear perception, the vehicle cannot continue to drive, as it may cause accidents. But once the vehicle stops in place, the driver needs to come and “rescue” it, and the experience of Intelligent Summon will become very poor. Users may not use it (just like many people almost don’t use automatic parking, not because there is no demand, but because the performance of the feature is too bad). Then the globally massive “user development” mode is completely ineffective.

The application scenarios of Intelligent Summon determine that its first version’s completion level must be very high, and Intelligent Summon itself is extremely challenging at the technical level. The combination of the two makes it not surprising that the release of this feature is delayed again. For Chinese car owners, since the complexity and compactness of parking lots in China are one order of magnitude higher than those in North America, everyone still needs to manage their expectations.To put it mildly, when intelligent summon is fully deployed around the world, the ability boundary of AP in high-speed scenarios, city public road scenarios, and even rural scenarios without lane lines will be completely another level.

Now, let’s talk about the things that have really been implemented in the driver assistance section.

Driving Visualization

This feature has been gradually released in the last version of V9, including vehicle support for dragging and zooming, better observation of the surrounding environment, and more traffic detection.

Automatic Lane Change

More intuitive automatic lane change animation has been updated, also in the last version of V9.

Map Improvement

The interactive logic of the map has been optimized, which is more convenient for navigation and supports quick viewing of destination details.

Sentry Mode Improvement

The Sentry mode can be saved in a separate folder for retrieval and management. When the Sentry mode occupies more than 5GB of space, the previous videos will be deleted automatically to release storage space.

Mainly, let’s talk about the updates for dragging and automatic lane change animation. Elon previously mentioned in an interview with Lex that Tesla pays great attention to driving visualization because it shows a world in a vector space, which not only helps drivers understand how strong AP is, but also how weak AP is. For example, in the early days, AP would recognize yaks as people, which means that AP’s ability to perceive large animals was extremely unreliable.

This is why Tesla actively updates the “Driving Visualization”.

Other Updates

Finally, there are some minor updates, and the only thing I want to mention is Joe mode.

Joe mode

When Joe mode is turned on, the warning sound of the vehicle will be reduced as much as possible to minimize the interference to passengers (especially children).

Software Update Optimization

Support for viewing update progress bars and updating versions in progress.

Underlying Menu Layout Optimization

After the release of V10, Elon suddenly tweeted that Joe mode is the best mode, and I’m actually looking forward to it. Tesla’s system warning sound is well done, but the downside is that the system is very sensitive to warns in advance. If the weight of the system is lowered, accidents may occur due to warnings coming too late.A suggestion from a user is to reduce the warning sound by half, thus avoiding disturbing users while maintaining its functionality. Considering that all of Elon’s five children have grown up, I guess he himself cannot stand frequent warning sounds. However, Joe should be the name of one of his children.

Finally, as of now, the Tesla Model 3 is the only vehicle in the industry that integrates vehicle operation (energy consumption, mileage), advanced driving assistance systems (ADAS visualization, maps), and information entertainment (games, videos, music) on the same screen.

Whether pursuing minimalist design or cost control, this design imposes higher requirements on the subsequent arrangement of information in a small space. Neither information entertainment nor advanced driving assistance is an easy job for Tesla, and there are many challenges ahead. Let’s see how Elon Musk responds.

Autopilot Progress from Elon Musk’s Twitter