据说还有空中悬浮障碍物，比如树枝，路上的栏杆。 前两天哪里看到austin robotaxi火车道口栏杆放下时候不停，安全员介入停下。 Grok: There is a reported incident involving a Tesla Robotaxi in Austin, Texas, where it nearly crossed a railway track as the crossing gate was lowering, signaling an approaching train. According to Joe Tegtmeyer, a Tesla enthusiast and YouTube host who was a passenger, the Robotaxi did not recognize the lowering gate and attempted to proceed onto the tracks. A Tesla safety supervisor in the passenger seat intervened to stop the vehicle, preventing a potential accident. The incident was not captured on video, and Joe downplayed its severity, noting the vehicle stopped at a traffic light before the tracks and did not actually cross them. He described the intervention as a cautious measure and rated the ride highly despite the issue. This event, reported on July 16, 2025, raised concerns about the Robotaxi’s ability to detect railway crossings, prompting scrutiny from critics and discussions about the need for software improvements. No other specific incidents of a Robotaxi running a railway crossing while the lever was lowering were found in the provided information.
大家也要注意一点，FSD 在不同的 Tesla hardware version上， 精确度，准确度， 电脑processing 速度上是有区别的。2023年以后的Tesla 用的是HW4，而之前的是HW3，HW4的camera resolution提高了，从1.3mp提高到5mp，加装了Radar,。 Key improvements and features of HW4: Enhanced processing power: HW4 includes a more powerful computer with three neural network processors, compared to HW3's two. Faster processing speed: HW4's computer, featuring an AMD Ryzen chip for infotainment, is significantly faster, resulting in a more responsive touchscreen and potentially faster FSD performance. Potential for second display: HW4 includes two additional, but depopulated, display connectors, hinting at a possible second display in future vehicles. Redesigned wiring and cooling: HW4 features a new wiring harness and cooling system to accommodate the more powerful computer and cameras, making retrofitting HW3 vehicles impractical. Improved cameras: HW4 utilizes higher-resolution cameras (5MP) compared to HW3's 1.2MP cameras, leading to better image quality and object detection. Tesla's Hardware 4 (HW4) vehicles are confirmed to include a new, high-definition radar system, despite Tesla previously moving away from radar in favor of camera-based systems. This new radar, code-named Phoenix, is expected to enhance situational awareness and improve the accuracy of Tesla's self-driving capabilities, especially in challenging conditions like low visibility. 所以我们驾驶的Tesla是HW3 还是HW4，大家可能有不同的体验。
比如carpool lane入口栏杆，铁路道口的栏杆， 似乎对空中悬浮障碍物识别有困难 读完grok的总结，似乎不是栏杆而是快速车道入口和铁路道口综合起来的复杂路况。 Tesla''s Full Self-Driving (FSD) system, which relies on camera-based vision and neural networks, has been reported to struggle with certain complex scenarios, including recognizing specific road features like levers or barriers at carpool lane entrances and railroad crossings. Here''s an analysis based on available information: Carpool Lane Entrances Challenges with HOV Lanes: There have been reports of Tesla''s FSD having difficulty navigating High-Occupancy Vehicle (HOV) or carpool lanes, particularly with entering or exiting these lanes. The issue often stems from FSD''s hesitation or inability to cross solid white lines, which are commonly used to designate HOV lanes. For example, a Tesla owner in Phoenix, AZ, noted that FSD struggles to merge into or out of HOV lanes, potentially due to the system''s conservative approach to solid line markings, which it may interpret as barriers. Levers or Barriers: While there is no direct mention in the provided references of FSD specifically failing to recognize "levers" at carpool lane entrances, the system may misinterpret or fail to act appropriately when encountering physical barriers, such as retractable levers or gates, if they are not clearly visible or if the system lacks specific training data for such objects. FSD''s reliance on visual cues means that unusual or less common road features, like levers, may not be consistently recognized, especially if they blend into the environment or are not part of the system''s trained dataset. Railroad Crossings Documented Issues: There are significant reports of FSD struggling at railroad crossings, particularly in recognizing active crossing signals, barriers, or moving trains. For instance, a Tesla owner in Ohio reported two incidents where their vehicle in FSD mode failed to detect a passing train, nearly resulting in collisions. In one case, the car did not slow down as it approached a railroad crossing with flashing lights and a moving train, only veering at the last moment, possibly due to driver intervention or an emergency response by the system. Crossing Arms (Levers): The same reports indicate that FSD failed to appropriately respond to railroad crossing arms (levers), with one incident resulting in the car hitting a crossing gate. The system''s failure to detect these barriers may be due to factors like foggy conditions, poor visibility, or the system''s inability to consistently interpret the crossing arm as an obstacle requiring a stop. Additional Evidence: Another report highlighted a case where FSD stopped on railroad tracks due to confusion caused by a stop sign immediately following the crossing, indicating that complex scenarios involving multiple signals (e.g., crossing arms and stop signs) can overwhelm the system''s decision-making process. Why These Issues Occur Vision-Based System: FSD relies entirely on cameras and neural networks (Tesla Vision) rather than radar or LIDAR, which can make it less effective in scenarios with low visibility (e.g., fog) or when objects like levers or crossing arms are not clearly distinguishable from the background. Training Data Limitations: FSD''s performance depends on the quality and diversity of its training data. Less common features, such as levers at carpool lane entrances or railroad crossing arms, may not be adequately represented in the training set, leading to recognition failures. Complex Scenarios: Both carpool lane entrances and railroad crossings often involve dynamic elements (e.g., moving trains, flashing lights, or temporary barriers) and require nuanced decision-making, which FSD is still developing. The system is in a supervised beta phase, meaning it requires active driver oversight and is not yet fully autonomous. Current Status and Recommendations Supervised Nature of FSD: Tesla emphasizes that FSD (Supervised) requires constant driver attention and intervention, especially in complex situations like railroad crossings or carpool lane transitions. Drivers must be prepared to take control if the system fails to recognize obstacles or make appropriate decisions. Software Updates: Tesla continuously improves FSD through over-the-air updates, which may address some of these issues over time. For example, updates have been issued to improve intersection handling, but railroad crossings and unique lane markings remain challenging. Driver Responsibility: Given the reported incidents, drivers should exercise extra caution at railroad crossings and carpool lane entrances, keeping hands on the wheel and being ready to intervene. Practicing disengaging FSD in safe environments is recommended to build familiarity with the process. Conclusion Yes, Tesla''s FSD has documented trouble recognizing and appropriately responding to levers or barriers at railroad crossings, as evidenced by multiple incidents where the system failed to detect crossing arms or trains. For carpool lane entrances, FSD struggles with navigating solid white lines and may not reliably handle physical levers or gates if present, though specific reports on levers are less common. These issues highlight the limitations of FSD''s current capabilities, particularly in complex or less common scenarios, and underscore the need for active driver supervision. Always stay alert and be prepared to take control, especially in such situations. If you have specific examples or locations in mind, I can look into further details or analyze related content if you provide more context
开长途fsd非常省力。
非常有必要 你不会后悔的
这种其实挺恐怖的，你不知道什么时候信任它，就是这时候出了问题怎么办？如果一直不好使，可能你还会提起精神看着它，但大多数时间都好用的话，可能就放松警惕反而更容易出事了。
值得的，我每天开车都用，很省劲，都不想自己开了。
据说还有空中悬浮障碍物，比如树枝，路上的栏杆。
前两天哪里看到austin robotaxi火车道口栏杆放下时候不停，安全员介入停下。
Grok:
There is a reported incident involving a Tesla Robotaxi in Austin, Texas, where it nearly crossed a railway track as the crossing gate was lowering, signaling an approaching train. According to Joe Tegtmeyer, a Tesla enthusiast and YouTube host who was a passenger, the Robotaxi did not recognize the lowering gate and attempted to proceed onto the tracks. A Tesla safety supervisor in the passenger seat intervened to stop the vehicle, preventing a potential accident. The incident was not captured on video, and Joe downplayed its severity, noting the vehicle stopped at a traffic light before the tracks and did not actually cross them. He described the intervention as a cautious measure and rated the ride highly despite the issue. This event, reported on July 16, 2025, raised concerns about the Robotaxi’s ability to detect railway crossings, prompting scrutiny from critics and discussions about the need for software improvements. No other specific incidents of a Robotaxi running a railway crossing while the lever was lowering were found in the provided information.
负责地说，v13这个版本出来以后就很靠谱了，现在根本不愿意自己开，fsd当司机用了，LA这种大街小巷没有任何问题，高速更是靠fsd。用了fsd 我的活动范围都大多了。
可以设置尽量不换lane
不恐怖，fsd也会一直提醒你让你注意监管。
注意$7500退税什么时候过期，大美丽法案刚刚砍掉这个退税了，不知道现在生效没有。
就是为了省传感器的钱，虽然传感器已经很便宜了 特斯拉的设计思想非常奇葩，一方面搞全玻璃天窗、无框门、隐藏门把手等成本高昂但是用处不大、弊端倒很多的东西，一方面把仪表盘、物理按钮拨杆、雨量传感器等成本不高但是很有用的东西省略掉，本质上就是PUA用户
特斯拉最不该省的是毫米波雷达和激光雷达，如果有，特斯拉涉及的事故大部分是可以避免的。现在毫米波雷达的成本不到100美元，激光雷达的成本不到1000美元。
雷达本身成本不高了，现在国内跟特斯拉同价位的EV都有好几个雷达，但是加入雷达后，自动驾驶的模型就要重新训练，训练后还不能兼容只有摄像头的老车型，这才是他一条道走到黑的原因。
我同事刚刚用fsd在路口撞了别人 幸好没大事
加入激光雷达后，自动驾驶的模型训练太复杂，基本不行。因为有些按光学传感器的，有些按激光雷达的，最后相互冲突。
那个判罚是因为特斯拉隐藏数据信息。
这真是一本正经的胡说八道了。国内所有做电车的公司都承认了纯视觉是唯一可行的解决方案。不能不佩服老马在技术上超前的直觉。
这不就是胡说八道吗，国内高档车清一色用激光雷达，有的还用两个或三个。
对，这个问题在不同地方场景看到特斯拉车主提到过。
高端的都要上相控阵了 特斯拉早期是因为雷达太贵 加上后售价太贵影响销量和股价 后来雷达便宜了 但是重新训练模型成本太高 只能一条道走到黑了
这个横杆子是什么东西？ 我目前的感受是乡下和高速都很好用，城里头大家开车没那么守规矩，Tesla不灵光，但是是那种比较安全的不灵，这怕那怕一只不开被后面的车滴。还有一个是控制自行车的红绿灯FSD识别不了是控制自行车的。所以我到城里几乎不用FSD
Key improvements and features of HW4: Enhanced processing power: HW4 includes a more powerful computer with three neural network processors, compared to HW3's two. Faster processing speed: HW4's computer, featuring an AMD Ryzen chip for infotainment, is significantly faster, resulting in a more responsive touchscreen and potentially faster FSD performance. Potential for second display: HW4 includes two additional, but depopulated, display connectors, hinting at a possible second display in future vehicles. Redesigned wiring and cooling: HW4 features a new wiring harness and cooling system to accommodate the more powerful computer and cameras, making retrofitting HW3 vehicles impractical. Improved cameras: HW4 utilizes higher-resolution cameras (5MP) compared to HW3's 1.2MP cameras, leading to better image quality and object detection.
Tesla's Hardware 4 (HW4) vehicles are confirmed to include a new, high-definition radar system, despite Tesla previously moving away from radar in favor of camera-based systems. This new radar, code-named Phoenix, is expected to enhance situational awareness and improve the accuracy of Tesla's self-driving capabilities, especially in challenging conditions like low visibility.
所以我们驾驶的Tesla是HW3 还是HW4，大家可能有不同的体验。
因人而异，我觉得好用，避免过两三次事故，坏处是自己的开车技能显著下降。
比如carpool lane入口栏杆，铁路道口的栏杆，
似乎对空中悬浮障碍物识别有困难
读完grok的总结，似乎不是栏杆而是快速车道入口和铁路道口综合起来的复杂路况。
Tesla''s Full Self-Driving (FSD) system, which relies on camera-based vision and neural networks, has been reported to struggle with certain complex scenarios, including recognizing specific road features like levers or barriers at carpool lane entrances and railroad crossings. Here''s an analysis based on available information:
Carpool Lane Entrances Challenges with HOV Lanes: There have been reports of Tesla''s FSD having difficulty navigating High-Occupancy Vehicle (HOV) or carpool lanes, particularly with entering or exiting these lanes. The issue often stems from FSD''s hesitation or inability to cross solid white lines, which are commonly used to designate HOV lanes. For example, a Tesla owner in Phoenix, AZ, noted that FSD struggles to merge into or out of HOV lanes, potentially due to the system''s conservative approach to solid line markings, which it may interpret as barriers. Levers or Barriers: While there is no direct mention in the provided references of FSD specifically failing to recognize "levers" at carpool lane entrances, the system may misinterpret or fail to act appropriately when encountering physical barriers, such as retractable levers or gates, if they are not clearly visible or if the system lacks specific training data for such objects. FSD''s reliance on visual cues means that unusual or less common road features, like levers, may not be consistently recognized, especially if they blend into the environment or are not part of the system''s trained dataset. Railroad Crossings Documented Issues: There are significant reports of FSD struggling at railroad crossings, particularly in recognizing active crossing signals, barriers, or moving trains. For instance, a Tesla owner in Ohio reported two incidents where their vehicle in FSD mode failed to detect a passing train, nearly resulting in collisions. In one case, the car did not slow down as it approached a railroad crossing with flashing lights and a moving train, only veering at the last moment, possibly due to driver intervention or an emergency response by the system. Crossing Arms (Levers): The same reports indicate that FSD failed to appropriately respond to railroad crossing arms (levers), with one incident resulting in the car hitting a crossing gate. The system''s failure to detect these barriers may be due to factors like foggy conditions, poor visibility, or the system''s inability to consistently interpret the crossing arm as an obstacle requiring a stop. Additional Evidence: Another report highlighted a case where FSD stopped on railroad tracks due to confusion caused by a stop sign immediately following the crossing, indicating that complex scenarios involving multiple signals (e.g., crossing arms and stop signs) can overwhelm the system''s decision-making process. Why These Issues Occur Vision-Based System: FSD relies entirely on cameras and neural networks (Tesla Vision) rather than radar or LIDAR, which can make it less effective in scenarios with low visibility (e.g., fog) or when objects like levers or crossing arms are not clearly distinguishable from the background. Training Data Limitations: FSD''s performance depends on the quality and diversity of its training data. Less common features, such as levers at carpool lane entrances or railroad crossing arms, may not be adequately represented in the training set, leading to recognition failures. Complex Scenarios: Both carpool lane entrances and railroad crossings often involve dynamic elements (e.g., moving trains, flashing lights, or temporary barriers) and require nuanced decision-making, which FSD is still developing. The system is in a supervised beta phase, meaning it requires active driver oversight and is not yet fully autonomous. Current Status and Recommendations Supervised Nature of FSD: Tesla emphasizes that FSD (Supervised) requires constant driver attention and intervention, especially in complex situations like railroad crossings or carpool lane transitions. Drivers must be prepared to take control if the system fails to recognize obstacles or make appropriate decisions. Software Updates: Tesla continuously improves FSD through over-the-air updates, which may address some of these issues over time. For example, updates have been issued to improve intersection handling, but railroad crossings and unique lane markings remain challenging. Driver Responsibility: Given the reported incidents, drivers should exercise extra caution at railroad crossings and carpool lane entrances, keeping hands on the wheel and being ready to intervene. Practicing disengaging FSD in safe environments is recommended to build familiarity with the process. Conclusion Yes, Tesla''s FSD has documented trouble recognizing and appropriately responding to levers or barriers at railroad crossings, as evidenced by multiple incidents where the system failed to detect crossing arms or trains. For carpool lane entrances, FSD struggles with navigating solid white lines and may not reliably handle physical levers or gates if present, though specific reports on levers are less common. These issues highlight the limitations of FSD''s current capabilities, particularly in complex or less common scenarios, and underscore the need for active driver supervision. Always stay alert and be prepared to take control, especially in such situations. If you have specific examples or locations in mind, I can look into further details or analyze related content if you provide more context