W08: AV Perception

Self-sensing

Uses sensors to measure current state of vehicle

Localisation

Uses sensors to determine vehicle's global and local position
Eg. Global Positioning System to determine geographic location

Surrounding-sensing

Uses sensors to sense surrounding

Eg. Hall effect sensor to measure vehicle velocity, Inertial Measurement Unit (IMU) to measure acceleration

Eg. Camera to provide road markings, LiDAR to detect obstacles

FAQ

Accuracy

The error between the true value and its measurement, which include noise levels and external interference rejection parameters

Resolution

The minimum difference between two measurements

Sensitivity

The smallest value that can be detected or measured

Dynamic Range

The minimum and maximum values that can be detected

Perspective

Quantities such as sensor range or its field of view

Active VS Passive

Whether the sensor emits energy or radiation that illuminates the environment or relies on ambient conditions

Timescale

Quantities such as update rate of sensor output and the frequency bandwidth of the measurement output over time

Output or interface technology

How the information is transmitted from the sensor to controller such as analogue voltage or current, digital outputs, serial or network data streams

Commonly available measurement on vehicles

Wheel speed

Produce a digital signal where its frequency is proportional to speed

Vehicle dynamic state

Include yaw rate, lateral and longitudinal acceleration

Driver inputs

Includes throttle and brake pedal positions, turn signals, headlights, steering wheel position

Brake pressure

Either at the master cylinder or for each wheel, usually measured by a diaphragm or silicone piezoelectric sensor
Silicon piezoelectric sensor measure changes in pressure, acceleration, strain, or force by converting them to electrical energy

Engine and exhaust variables

Includes coolant temperature, engine speed, O2 and NOx levels, spark plug firing timing

Global Navigation Satellite System (GNSS)

uses satellites to determine location

Uses time signals transmitted along a line of sight by radio from multiple satellites

Information available from a receiver

  • Absolute position
  • Velocity and orientation
  • Precise time and synchronised pulse per second
  • Raw information that can be used for precise post-processing applications

Uses sensors to perceive road markings, road slope, traffic signs, weather conditions, the state (position, velocity, acceleration, etc.)

Advantages

  • Cheap
  • Work normally in low visibility and low light conditions

Disadvantages

  • Can only be used at low speed, for example during parking
  • Shorter field of view and accuracy compared to LiDAR

Image processing sensor

Advantages

  • Low-cost
  • Implementation is almost entirely in software

Disadvantages

  • Implemented as passive sensors, thus must cope with full range of ambient and uncontrolled conditions
  • Requires a lot of processing power from the controller, which also increase heat generation