ADAS sensors and functions: radar, camera, lidar

How ADAS functions use radar (FMCW and Doppler), cameras, lidar and ultrasonic sensors, their strengths and limits, SAE automation levels and sensor fusion, with time-of-flight, Doppler, FMCW, time-to-collision and AEB stopping-distance numericals.

Drafted with Aria, reviewed by the AiCanCode.org team. Spotted an error? Use Give Feedback at the bottom of the page.

Why it matters

Advanced driver-assistance systems (ADAS) — automatic emergency braking, adaptive cruise control, lane-keeping, blind-spot warning, parking aids — are spreading quickly through Indian cars, and safety ratings and regulations increasingly expect them. Each function is only as good as the sensors behind it. An automotive engineer must know what radar, cameras, lidar and ultrasonic sensors can and cannot see, how ranges and speeds are measured, and how timing and braking physics decide whether a collision is avoided.

Key ideas

Levels of automation (SAE J3016). Level 0 warnings only; Level 1 one function assists (steering or speed, e.g. ACC); Level 2 combined steering and speed with the driver supervising; Level 3 the system drives in defined conditions and the driver must take over on request; Levels 4–5 no driver needed within (L4) or beyond (L5) a defined domain. Almost all ADAS on sale is Level 1–2: the driver remains responsible.

Typical functions. Forward-collision warning and autonomous emergency braking (AEB), adaptive cruise control (ACC) with a set time gap, lane-departure warning and lane-keep assist, blind-spot detection, rear cross-traffic alert, traffic-sign recognition, automatic high beam, park assist and surround view, driver-monitoring cameras.

Radar.

  • Automotive radars use 76–81 GHz (77 GHz band; 24 GHz in older short-range units). Most use FMCW: the transmitted frequency ramps (a "chirp"), the echo returns delayed, and mixing gives a beat frequency proportional to range. The Doppler shift gives the relative (radial) speed directly; an antenna array gives the angle.
  • Long-range radar sees about 200–250 m in a narrow beam (ACC, AEB); short- and mid-range radars cover wide angles near the car (blind spot, cross traffic).
  • Strengths: works in darkness, rain, fog and dust; measures speed directly. Weaknesses: limited angular resolution, poor at classifying objects, false returns from metal structures, and stationary objects are hard to separate from roadside clutter.

Camera.

  • Mono or stereo cameras behind the windscreen, plus surround-view and rear cameras. Image processing and neural networks detect lane markings, signs, vehicles, pedestrians and cyclists and classify them — something radar cannot do well.
  • Mono cameras estimate distance from object size and position; stereo cameras compute depth from the disparity between two images.
  • Weaknesses: needs light and a clean windscreen; affected by glare, low sun, darkness, heavy rain and fog; large processing load.

Lidar.

  • Emits laser pulses (typically 905 nm or 1550 nm) and times their return, scanning to build a 3-D point cloud with fine angular resolution, giving accurate shape and position.
  • Weaknesses: degraded by fog, heavy rain, snow and dirt on the window; higher cost; mainly used in higher automation levels.

Ultrasonic sensors. 40–50 kHz transducers in the bumpers measuring echo time over about 0.2–5 m for parking; cheap, but short range and slow.

Sensor fusion and the control chain. Data from several sensors are combined (e.g. radar range and speed with camera classification) to reduce false alarms and missed detections. The chain is sense → perceive (detect, track, classify) → decide (e.g. time to collision) → act (warning, brake pressure via ESC, steering torque via electric power steering). System latency and the available deceleration decide what can be avoided.

Calibration. Radars and cameras must be aligned after windscreen replacement, bumper repair or wheel-alignment changes; a misaligned sensor can brake for the wrong object or miss the right one.

Formulas

d = v_w × t / 2

  • d: distance to target (m), v_w: wave speed (m/s; 3 × 10⁸ for radar and lidar, about 343 for ultrasound in air at 20 °C), t: round-trip time (s). Divide by 2 because the wave goes out and back.

f_D = 2 v_r / λ, λ = c / f

  • f_D: Doppler shift (Hz), v_r: closing (radial) speed (m/s), λ: wavelength (m), f: carrier frequency (Hz).

ΔR = c / (2B), f_b = 2 R B / (c T_c)

  • FMCW radar: ΔR: range resolution (m), B: chirp bandwidth (Hz), f_b: beat frequency (Hz), R: range (m), T_c: chirp duration (s).

Z = f × b / disparity

  • Stereo camera depth: Z (m), f: focal length (pixels), b: baseline between cameras (m), disparity (pixels).

TTC = d / v_rel

  • Time to collision (s) at constant closing speed v_rel (m/s).

s_stop = v × t_lat + v² / (2a)

  • s_stop: distance to stop (m), t_lat: total sensing-plus-actuation latency (s), a: deceleration (m/s²).

d_ACC = d₀ + v × t_gap

  • ACC desired gap (m), d₀: standstill distance (m), t_gap: set time gap (s).

Worked examples

Example 1 (standard). (a) A radar echo returns 1.0 µs after transmission. Find the range. (b) A parking sensor's echo returns after 5.83 ms. Find the distance. (c) A 77 GHz radar sees a car closing at 20 m/s. Find the Doppler shift.

  1. d = 3 × 10⁸ × 1.0 × 10⁻⁶ / 2 = 150 m.
  2. d = 343 × 5.83 × 10⁻³ / 2 = 1.00 m.
  3. λ = 3 × 10⁸ / 77 × 10⁹ = 3.90 × 10⁻³ m; f_D = 2 × 20 / 3.90 × 10⁻³ = 10.3 × 10³ Hz.

Answer: 150 m, 1.00 m, about 10.3 kHz.

Example 2 (GATE level). A car at 72 km/h approaches a stationary vehicle. Its radar first detects the obstacle at 60 m. The AEB system has a total latency of 0.3 s and then brakes at 8 m/s². (a) Find the TTC at detection. (b) Will it stop in time, and with what margin? (c) What is the last distance at which AEB could still trigger and avoid impact? (d) The radar uses a 1 GHz chirp over 50 µs. Find its range resolution and the beat frequency for a target at 60 m.

  1. v = 72 / 3.6 = 20 m/s; TTC = 60 / 20 = 3.0 s.
  2. Latency distance = 20 × 0.3 = 6 m; braking distance = 20² / (2 × 8) = 25 m; s_stop = 31 m.
  3. Margin = 60 − 31 = 29 m, so the collision is avoided. The last trigger distance is 31 m (TTC = 31 / 20 = 1.55 s).
  4. ΔR = 3 × 10⁸ / (2 × 10⁹) = 0.15 m; f_b = 2 × 60 × 10⁹ / (3 × 10⁸ × 50 × 10⁻⁶) = 8.0 × 10⁶ Hz.

Answer: TTC 3.0 s; stops 29 m short; last trigger at 31 m (1.55 s TTC); 0.15 m resolution and 8 MHz beat frequency.

Common mistakes

  • Forgetting the factor 2 for the round trip in time-of-flight calculations.
  • Using the speed of light for ultrasonic sensors, or 343 m/s for radar.
  • Quoting radar or lidar ranges of kilometres; automotive sensors work over metres to about 250 m, so round-trip times are nanoseconds to about 2 µs.
  • Assuming radar measures sideways (tangential) speed; Doppler gives only the radial component.
  • Converting km/h to m/s incorrectly (divide by 3.6).
  • Ignoring latency in AEB calculations — at 20 m/s, 0.3 s is 6 m.
  • Treating a Level 2 system as self-driving; the driver must supervise at all times.

For GATE ME

ADAS questions reduce to kinematics and wave basics: time of flight, Doppler shift, relative motion, stopping distance with reaction time, and time to collision. Practise unit conversion (km/h, m/s, µs, GHz) and the constant-deceleration equations.

Quick check

  1. Why does radar outperform cameras in fog?
  2. What does the Doppler shift tell an ADAS radar?
  3. A lidar pulse returns after 400 ns. How far away is the target?
  4. Which sensor is best at reading traffic signs?
  5. A car at 90 km/h is 50 m behind a stationary obstacle. What is the TTC?

Answers: 1. Millimetre-wave radio is much less scattered by fog droplets than visible light. 2. The radial relative speed of the target. 3. 3 × 10⁸ × 400 × 10⁻⁹ / 2 = 60 m. 4. The camera. 5. 50 / 25 = 2.0 s.

Interview questions

All Automotive Electronics and Electric Vehicles interview questions

Try answering each one aloud before you open it.

  1. 1.What is ADAS and what are its primary functions in vehicles?Concept

    ADAS stands for Advanced Driver Assistance Systems. It is a collection of electronic technologies that assist drivers in driving and parking functions. The primary functions of ADAS include enhancing vehicle safety and road safety by providing features such as adaptive cruise control, lane departure warning, collision avoidance, and parking assistance.

  2. 2.Explain how radar sensors work in ADAS.Concept

    Automotive radars transmit at about 77 GHz, mostly as FMCW, where the frequency is swept in a ramp or chirp. The echo returns delayed, and mixing it with the transmitted signal gives a beat frequency proportional to range. The Doppler shift, f_D = 2v_r/λ, gives the target's radial relative speed directly, and an array of receive antennas gives its angle. Long-range radar sees about 200–250 m for ACC and AEB, and it keeps working in darkness, rain and fog, but its angular resolution and ability to classify objects are limited.

  3. 3.Describe the role of cameras in ADAS.Concept

    Cameras in ADAS are used to capture visual information from the vehicle's surroundings. They provide data for features like lane departure warning, traffic sign recognition, and pedestrian detection. Cameras can process images to identify lane markings, road signs, and other vehicles, helping the system make informed decisions to assist the driver.

  4. 4.What is LiDAR and how is it used in ADAS?Concept

    LiDAR, which stands for Light Detection and Ranging, uses laser beams to create a 3D map of the vehicle's surroundings. It emits laser pulses that reflect off objects and return to the sensor. By measuring the time it takes for the pulses to return, LiDAR can determine the distance and shape of objects. In ADAS, LiDAR is used for precise object detection and mapping, which is crucial for autonomous driving features.

  5. 5.Why is radar preferred over LiDAR in certain ADAS applications?Application

    Radar is preferred for ACC and AEB because millimetre waves pass through rain, fog, dust and darkness far better than laser light, it measures relative speed directly through Doppler, and it is much cheaper and easier to hide behind a bumper. A long-range radar reaches about 200–250 m. Lidar gives much finer angular resolution and an accurate 3-D shape of objects, which is why it is used for mapping and higher levels of automation, but it is costlier and degraded by fog, rain, snow and dirt.

  6. 6.What happens if a camera sensor in an ADAS system fails?Application

    If a camera sensor in an ADAS system fails, the features that rely on visual data, such as lane departure warning, traffic sign recognition, and pedestrian detection, may become inoperative. The system may alert the driver with a warning message or sound, indicating that the affected features are not available. It is crucial for the driver to be aware of this and take manual control as needed.

  7. 7.How do ADAS systems integrate data from radar, camera, and LiDAR sensors?Application

    ADAS systems integrate data from radar, camera, and LiDAR sensors through a process called sensor fusion. Sensor fusion combines the data from different sensors to create a comprehensive understanding of the vehicle's environment. This integration allows the system to compensate for the limitations of individual sensors and make more accurate and reliable decisions for driver assistance.

  8. 8.A radar signal returns 1.0 µs after it was transmitted. Taking the speed of radio waves as 3 × 10⁸ m/s, how far away is the object?Numerical

    The wave travels to the object and back, so d = c × t / 2 = 3 × 10⁸ × 1.0 × 10⁻⁶ / 2 = 150 m. That is near the useful range of a long-range automotive radar of about 200–250 m. Typical automotive round-trip times are therefore well under 2 µs; delays of milliseconds would correspond to hundreds of kilometres.

  9. 9.A lidar pulse returns 400 ns after it was emitted. Taking the speed of light as 3 × 10⁸ m/s, what is the distance to the object?Numerical

    Using the round-trip relation, d = c × t / 2 = 3 × 10⁸ × 400 × 10⁻⁹ / 2 = 60 m. Lidar times are measured in nanoseconds: 1 ns of timing error equals 15 cm of range, which is why lidar needs very fast timing electronics.

  10. 10.Explain the limitations of using cameras in ADAS systems.Application

    Cameras in ADAS systems have limitations such as being affected by poor lighting conditions, glare, and weather conditions like fog and rain. They may struggle to accurately detect objects in low light or when the sun is directly in the camera's field of view. Additionally, cameras require significant processing power to analyze images and can be limited by their field of view.

Finished this topic? Mark it so your progress, study plan and readiness keep up.

Stuck on something here?

PreviousIn-vehicle networks: CAN, LIN and FlexRayNext Electric vehicle architecture and drivetrain sizing