來源:
05 Uber's Self-Driving Technology
Base Vihicle:Volvo XC90 (be recognized as one of the safest vehicls in the world)
Important features of vehiicle:
AEB: includes a forwarding facing radar + camera + ECU(Elecctronic Control Unit) .
The AEB operaatesindependently of Uber's self-driving system.
ABS : help to prevent lockup.
Seatbelt with Pre-tensioners and Load Limiters :
Pre-tensioners tighten safety belts in the event of collision , load limiters minimize belt-inflicted injury.
ESC: consists of traction control (牽引力控制) , spin control (旋轉控制) ,active yaw control (主動偏航控制), and engine drag control (發動機拖曳控制)
reduce wheel spin, counteract skidding, and improve directional stability. (減少車輪旋轉,抵消打滑,提高方向穩定性)
Hardware(Current Generationn)
1. Lidar
Num: 1, top-mounted
Range: over 100m
2. Camera
Range:high resolution (高分辨率), near-range(5m from vehicle), medium-range, and long-range imagery.
View: parts have a wild FOV(field of view), some haave narrow FOV.
Installation location:mounted in the sensor pod on ttop of the vehicle and around the vehicle for 360°
3. Radar
Foward-facing radars: mounted below the headlamps;
Side-facing radars: mounted in the front and rear corners of the vehicle;
Rear-facing radars: mounted near the ends of the bumper beam(保險槓樑).
4. GPS(Global Position System)
support localization
5. Self-driving Computer
run Perception, Predictiion, Mottion Planning, and other software.
Hardware and firmware are custom.
be liquid-dooled.
6. Telematics
Custom telematics hardware snd software provide cellular data communiication to support carrier network reduncancy, ssecureemobile data traffic, and authenticated cloud communication.
Software(Next Generation)
1. Ultrasonic Sensors (USSs)
USS provide near-ranging sensing of people he oobjects witnin 5m from the vihicle, in particular to assist with stopping & starting, lane changing, parallel parking.
優勢:短距離檢測行人和目標,尤其是停車/啓動/變道和停車入庫。
2. Vehicle Interface Module(VIM)
It is developed according to ISO26262 ASIL-D, provides closed-loop motion control, undertaking both trajectory management and trajectory tracking.
be fully redundant
3. High-Definition Map(HD Map) 高清地圖
Information layers included in Uber's HD Map:
- Geometry of the road and curbs;
- Drivable surface boundaries and driveways;
- Lane boundaries, including print llines of variious types;
- Bike and bus lanes, parking regions, stop lines, crosswalkers;
- Traffic control signals, llight sets, and lane and conflict associations;
- Railroad crossing and trolley or railcar tracks;
- Speed limits, constraints zones, restricctions, speed bumps;
- Traffic control signage;
4. automated tasks performed by self-driving software:
1. Perception || 感知:周邊環境建模
The Perception software processes the data gathered by overlapping sensors coving 360° and combine it with map intto a full representation of the environment.
Perception Software 傳感器蒐集的車身周圍信息和高精度地圖結合,真實呈現周圍環境。
2. Localization || 定位:知道自己在哪
determine where the vehicle is in the world, down to within a few centimeters.
3. Prediction || 預測:預測周圍目標要做什麼
prediict what the actors or objects in the invironment are likely to do next.
4. Routing & Navigation || 路線規劃 & 導航:選擇路線
determine what route the vehhicle should use to reach its intended destination.
5. Motion Planning || 動作規劃
motion plan for the vehicle.
6. Vehicle Control 車輛監控
06 Uber's Self-Driving Safety Principles
Safety Case
6.1 Proficient
6.2 Fail-safe
6.3 Continuously Improving
6.4 Resilient
6.5 Trustworthy
PS:
AEB: Automated Emergency Brakking 自動緊急剎車系統
ABS: Anti-locckk Breaking System 防抱死系統
ESC: Electronic Stability Control 電子穩定控制系統 各家叫法不同,e.g. ESP (大衆), VSA(本田 ),VCS(豐田)