Non-contact Sensing: Ultrasonic, LIDAR and RADAR — Robotics Tutorials

Ultrasonic Sensing: Echolocation for Robots

Ultrasonic sensors measure distance using the same echolocation principle as bats: a piezoelectric transducer emits a burst of high-frequency sound (typically 20 kHz to 200 kHz) and then switches to receive mode, measuring the elapsed time until the echo returns from a nearby surface.

Operating principle: The transducer applies an alternating voltage to a piezoelectric crystal, causing it to vibrate at resonance and emit a pressure wave. After the transmission burst, the same (or a paired) transducer detects the reflected acoustic energy. A timing circuit measures the time-of-flight from emission to echo detection.

Practical characteristics:

Range: Typically 2 cm to 10 m (depending on model); closer than ~2 cm creates a "blind zone" where transmission and reception overlap
Beam width: 15–30° cone angle; the sensor reports the nearest object within this cone, providing no directional discrimination within the beam
Update rate: 10–50 Hz (limited by sound travel time and decay)
Temperature dependence: Sound velocity varies with temperature (≈ 0.6 m/s per °C); compensation is required for accurate ranging over wide temperature ranges
Surface angle sensitivity: Surfaces at large angles to the beam may scatter sound away and produce no echo (specular reflection loss)
Cost: $0.50 – $50 (extremely inexpensive)

Common robotics examples include the HC-SR04 (hobby robotics), Maxbotix sensors (indoor mobile robots), and automotive parking sensors (48 kHz).

Ultrasonic Distance Calculation

Distance is computed from the time-of-flight of the acoustic pulse:

d = (v_sound × t_flight) / 2

where:
  d = distance to object (m)
  v_sound = speed of sound (m/s) ≈ 343 m/s at 20°C
  t_flight = round-trip travel time (s)
  Factor of 2: the sound travels to the object and back

Temperature correction:
  v_sound(T) = 331.3 + 0.606 × T  [m/s, T in °C]

For t_flight = 5.83 ms:
  d = (343 × 0.00583) / 2 = 1.0 m

Without temperature compensation, a 20°C error introduces ~3.5% range error — approximately 3.5 cm error at 1 m range.

LIDAR: Precision 3D Environment Mapping

LIDAR (Light Detection And Ranging) measures distance using laser pulses, producing highly accurate range measurements and, in 3D systems, dense point clouds representing the geometry of the environment.

Time-of-flight (ToF) LIDAR: A laser diode emits a nanosecond pulse. A photodiode (avalanche photodiode or SPAD — Single-Photon Avalanche Diode) detects the return pulse. The time interval is measured with picosecond-resolution counters. Since light travels at c ≈ 3×10⁸ m/s, measuring sub-centimeter range requires timing resolution on the order of 30 picoseconds.

Continuous-wave FMCW LIDAR: A frequency-modulated continuous wave approach measures range through the frequency difference between transmitted and received beams. This also enables direct velocity measurement via the Doppler shift — emerging in automotive applications.

Scanning mechanisms:

Mechanical rotating LIDAR: A mirror or prism spins at 10–20 Hz, sweeping a laser beam through 360°. Multiple laser channels at different vertical angles produce a 3D scan. Examples: Velodyne HDL-64E (64 channels), Ouster OS1-128.
Solid-state LIDAR: No moving parts — uses MEMS mirrors, optical phased arrays, or flash illumination. More compact and reliable; examples: Livox Mid-360, Luminar Iris (automotive).
2D scanning LIDAR: A single-layer 180°–360° scan in a horizontal plane. Classic example: SICK LMS291, Hokuyo URG-04LX — widely used in ROS navigation for obstacle detection.

LIDAR specifications vary widely: range 10–300 m, angular resolution 0.1°–0.4°, range accuracy ±2–5 cm, point rate 100,000–4,000,000 points/second.

LIDAR Range Equation

LIDAR range is computed from the speed of light and the round-trip travel time:

d = (c × Δt) / 2

where:
  d = distance to target (m)
  c = speed of light = 2.998 × 10⁸ m/s
  Δt = round-trip time-of-flight (s)

For d = 100 m:
  Δt = 2 × 100 / (2.998 × 10⁸) = 667 ns

Timing resolution needed for 1 cm range accuracy:
  Δt_min = 2 × 0.01 / c = 66.7 ps  (picoseconds)

The LIDAR range equation is identical in form to the ultrasonic equation, but with c (speed of light) instead of v_sound — hence LIDAR requires picosecond-resolution timing whereas ultrasonic needs only microsecond resolution.

RADAR: All-Weather Long-Range Sensing

RADAR (Radio Detection And Ranging) uses radio-frequency electromagnetic waves (typically 24 GHz for short-range or 77 GHz for automotive long-range) to detect objects and measure their range and velocity.

Key RADAR advantages for robotics:

Weather penetration: Radio waves penetrate fog, rain, dust, and darkness — where cameras and LIDAR degrade significantly
Velocity measurement: The Doppler effect enables direct measurement of the radial velocity of detected objects without differentiation, providing instantaneous velocity with high accuracy
Long range: Automotive RADAR detects objects at 200 m+ with consistent reliability
Low cost at scale: Automotive demand has driven 77 GHz RADAR modules below $50 in volume

FMCW (Frequency-Modulated Continuous Wave) RADAR is the dominant automotive architecture: a chirp signal is transmitted and the received signal is mixed with the transmitted signal. The beat frequency encodes range; the Doppler frequency encodes velocity; angle-of-arrival processing (multiple receive antennas) provides azimuth estimation.

Limitations for robotics:

Low spatial resolution compared to cameras and LIDAR — typically 1°–5° angular resolution
Strong returns from metallic surfaces; weak returns from pedestrians and soft materials
Multipath reflections in cluttered indoor environments cause ghost detections
Range ambiguity: radar cross-section (RCS) varies enormously across target types

Non-Contact Sensing Technology Comparison

Technology	Range	Angular Resolution	Frame Rate	Weather Affected	Velocity Meas.	Typical Cost
Ultrasonic	0.02 – 10 m	15 – 30° (beam width)	10 – 50 Hz	Minimal (wind affects)	No (indirect)	$1 – $50
2D LIDAR	0.1 – 80 m	0.1 – 0.5°	10 – 40 Hz	Rain/fog reduce range	No	$200 – $3,000
3D LIDAR (mechanical)	0.1 – 200 m	0.1 – 0.4° horizontal, 0.1 – 2° vertical	10 – 20 Hz (full scan)	Rain/fog reduce range	No (ToF)	$500 – $50,000
3D LIDAR (solid-state)	0.1 – 150 m	0.05 – 0.2°	10 – 25 Hz	Rain/fog reduce range	Yes (FMCW)	$200 – $5,000
Automotive RADAR (77 GHz)	0.2 – 250 m	1 – 5°	10 – 100 Hz	Minimal (penetrates fog/rain)	Yes (Doppler)	$20 – $500
mmWave RADAR (60 GHz indoor)	0.1 – 10 m	2 – 5°	10 – 50 Hz	Minimal	Yes (Doppler)	$30 – $300

LIDAR vs. RADAR for Autonomous Vehicles

LIDAR

High-resolution 3D point clouds (millions of points/sec)
Sub-centimeter range accuracy (±1–3 cm)
Rich geometric detail — lane markings, object shapes
Degraded by rain, fog, snow, and dust
Eye-safe wavelengths require careful power management
Cannot directly measure target velocity — requires frame differencing
Moving to solid-state designs for reliability and cost reduction
Essential for HD map creation and precise localization

RADAR

Low spatial resolution — cannot distinguish fine geometry
All-weather operation — minimal degradation in rain/fog/snow
Direct Doppler velocity measurement — critical for traffic speed estimation
Long range (200 m+) for highway safety applications
Low cost at automotive scale — standard equipment on modern cars
Cannot image lane markings or identify object types by shape
Complementary to LIDAR and cameras, not a replacement
4D imaging RADAR (range, azimuth, elevation, Doppler) emerging

Key terms

LIDAR: Light Detection And Ranging — a sensor that measures distances using pulsed or continuous-wave laser beams, producing 2D scans or dense 3D point clouds of the environment.
Time-of-flight (ToF): A ranging principle that measures the elapsed time for a signal (light, sound, or radio wave) to travel to a target and return, then computes distance using the known signal velocity.
Ultrasonic transducer: A piezoelectric device that converts electrical energy to ultrasonic sound waves (and vice versa), used for proximity sensing and distance measurement in robotics.
RADAR: Radio Detection And Ranging — a sensor using radio-frequency electromagnetic waves to detect objects and measure their range, angle, and radial velocity.
Doppler effect: The change in frequency of a wave (sound, light, radio) as observed when the source and observer have relative motion; used in RADAR and FMCW LIDAR to directly measure target velocity.
Point cloud: A set of data points in 3D space, each representing a measured surface location; the primary output format of 3D LIDAR sensors.
3D mapping: The construction of a three-dimensional geometric representation of an environment from sensor data, typically using LIDAR point clouds combined with robot pose estimates.
Solid-state LIDAR: A LIDAR design with no mechanically rotating components, using MEMS mirrors, optical phased arrays, or flash illumination for beam steering; more compact and reliable than mechanical spinning LIDAR.
Beam divergence: The angular spread of a laser or acoustic beam as it propagates; wider divergence reduces spatial resolution at long range but increases coverage area.
FMCW (Frequency-Modulated Continuous Wave): A ranging technique using a continuously transmitted signal with linearly varying frequency; the beat frequency between transmitted and received signals encodes both range and velocity.

← Sensing & Perception