当前位置:   article > 正文

Tesla FSD 自动驾驶软件升级版本说明书解析一(Beta v10.11.2 Release Notes)_tesla fsd beta 软件功能

tesla fsd beta 软件功能

       Tesla FSD软件的版本说明书写得十分技术流,里面会详细列举新增了什么feature、修复了什么bug、采用了什么方法(算法还是数据等)提升了多少指标之类的信息,有助于从业人员了解Tesla的底层算法技术和功能设计逻辑,值得阅读研究。

       Telsa FSD于2022年4月4日正式发布了Beta v10.11.2版本,适用于Model S/Model 3/Model X/Model Y。Release notes一共列写了13条如下,让我们逐条解析看看:

1. Upgraded modeling of lane geometry from dense rasters ("bag of points") to an autoregressive decoder that directly predicts and connects "vector space" lanes point by point using a transformer neural network. This enables us to predict crossing lanes, allows computationally cheaper and less error prone post-processing, and paves the way for predicting many other signals and their relationships jointly and end-to-end.

将车道拓扑结构关系建模从密集栅格(点阵)升级为自回归解码器,使用transformer神经网络直接预测并逐点联接“vector space(向量空间)”中的车道。这样能够直接预测交叉车道,降低了运算量并减少了原先基于规则的后处理算法的错误率,同时为后续做端到端的联合预测其他信号及其之间的关系铺平了道路。

2. Use more accurate predictions of where vehicles are turning or merging to reduce unnecessary slowdowns for vehicles that will not cross our path.

更加准确地预测出周边车辆可能进行转向或者并道的位置,从而降低自车因错误避让决策而产生的不必要刹车。

3. Improved right-of-way understanding if the map is inaccurate or the car cannot follow the navigation. In particular, modeling intersection extents is now entirely based on network predictions and no longer uses map-based heuristics.

在地图不准确或者车辆无法跟随导航场景下,提升了对车辆通行权的理解能力。特别地,对于路口区域拓扑的建模完全采用深度学习的方法而不再使用基于地图的启发式方法。

4. Improved the precision of VRU detections by 44.9%, dramatically reducing spurious false positive pedestrians and bicycles (especially around tar seams, skid marks, and rain drops). This was accomplished by increasing the data size of the next-gen autolabeler, training network parameters that were previously frozen, and modifying the network loss functions. We find that this decreases the incidence of VRU-related false slowdowns.

VRU(Vulnerable Road User,弱势道路使用者)的精度提升了44.9%!显著降低了行人和骑行人的误报(特别是由于柏油路裂缝、刹车痕迹、雨天等因素造成的)。这主要得益于三方面的优化:使用了下一代自动化标注系统增加了训练数据量、优化了任务loss、使用原先训好的网络进行finetuning。总之,能够降低与VRU相关的误刹车行为,提升舒适度。

5. Reduced the predicted velocity error of very close-by motorcycles, scooters, wheelchairs, and pedestrians by 63.6%. To do this, we introduced a new dataset of simulated adversarial high speed VRU interactions. This update improves autopilot control around fast-moving and cutting-in VRUs.

针对近处的摩托车、滑板车、轮椅、行人等VRU目标,速度预测误差降低了63.6!为此,我们构建了一个新的数据集,包含各类模拟对抗性的快速交互的VUR目标。该优化使得Autopilot系统能够更加轻松应对周边快速移动和突然插入的VRU。

6. Improved creeping profile with higher jerk when creeping starts and ends.

提高了“蠕行模式/怠速模式”启动和结束时的加加速度。

传统的燃油车能通过怠速让汽车在刹车和油门踏板松开的时候向前滑行,而电机完全可以没有怠速状态。Tesla为了照顾司机习惯而设置了“creep mode”来模拟怠速功能,这样在停车场和小巷之类狭窄的地方低速方便控制。

7. Improved control for nearby obstacles by predicting continuous distance to static geometry with the general static obstacle network.

使用通用静态障碍物检测网络连续预测静态障碍物的距离信息,提升了近处此类障碍物场景下的控制表现。

8. Reduced vehicle "parked" attribute error rate by 17%, achieved by increasing the dataset size by 14%. Also improved brake light accuracy.

通过增加14%的训练数据,对车辆“是否处于停车状态”(动静判断)属性预测错误率降低17%!同时对他车刹车灯的预测准确率也有提升。

9. Improved clear-to-go scenario velocity error by 5% and highway scenario velocity error by 10%, achieved by tuning loss function targeted at improving performance in difficult scenarios.

通过针对困难场景调整任务loss,提高了他车速度预测精度,其中转弯场景(clear-to-go)提高5%,高速场景提高10%。

10. Improved detection and control for open car doors.

提升了“他车车门是否打开”的检测精度和控制。Mobileye过往介绍中也有类似功能,对于避免碰撞比较重要。

11. Improved smoothness through turns by using an optimization-based approach to decide which road lines are irrelevant for control given lateral and longitudinal acceleration and jerk limits as well as vehicle kinematics.

使用基于优化的方法提升转弯时的平稳性,具体地使用该方法识别出哪些车道对车辆控制无关紧要,在给定横纵向加速度和加加速度及车辆运动学的情况下。

12. Improved stability of the FSD Ul visualizations by optimizing ethernet data transfer pipeline by 15%.

通过优化网络传输管道15%,增加了FSD UI可视化功能稳定性。

13. Improved recall for vehicles directly behind ego, and improved precision for vehicle detection network.

提高了车辆检测网络的精度,以及自车身后他车检测的召回率。

综上所述我们可以看到此次升级涉及到的模块归纳如下:

道路结构认知:1、3

感知(perception):4(VRU检测)、5(VRU速度)、7(静态障碍物测距)、8(动静判断)、9(车门检测)、10(车辆速度)、13(车辆检测)

预测(prediction):2

规控:6、11

其他:12

提及“仿真”:5

提及“自动化标注”:4

知乎:Tesla FSD 自动驾驶软件升级版本说明书解析一(Beta v10.11.2 Release Notes) - 知乎

知识星球:知识星球 | 深度连接铁杆粉丝,运营高品质社群,知识变现的工具

声明:本文内容由网友自发贡献,不代表【wpsshop博客】立场,版权归原作者所有,本站不承担相应法律责任。如您发现有侵权的内容,请联系我们。转载请注明出处:https://www.wpsshop.cn/w/小蓝xlanll/article/detail/544299
推荐阅读
相关标签
  

闽ICP备14008679号