Radar, ultrasonic, encoders, force/torque, tactile, UWB and sonar

1. Engineering problem

These sensors rarely carry a robot alone but often decide whether it works in practice: radar in weather, ultrasonic near glass, encoders for odometry and joint control, force/torque and tactile for contact, UWB for indoor positioning and sonar underwater.

2. When is this needed?

• Choosing a sensor type for a new robot, or a replacement when one fails to meet a requirement.
• Explaining to a team why a sensor behaves differently in the field than on the bench.
• Writing the datasheet checklist before you buy.

3. Selection and design decisions

Every sensor type below follows the same 18-point template, from what it measures to the alternatives. Use the interactive Sensor Comparison Tool to compare any of them side by side.

4. Advantages and disadvantages

• Radar vs LiDAR: radar works in rain and fog and measures velocity but is coarse; LiDAR is fine-grained but weather-limited.
• Ultrasonic: trivial and cheap, wide beam and short range.
• Encoders: excellent but measure the wheel, not the ground.
• Force/torque vs tactile: F/T measures total wrench; tactile measures local contact; both are expensive and fragile.
• UWB: needs anchors and surveying; accuracy depends on geometry.

5. Requirements and constraints

• Regulatory approval for radar and UWB frequency bands in your region.
• Mounting behind materials transparent to the band (radar) or with a clear acoustic path (ultrasonic, sonar).
• Cable strain relief at moving joints (F/T, tactile).
• Interfaces and driver availability: several have vendor-specific messages.

6. Architecture

Each sensor feeds the estimator or controller it supports: radar and ultrasonic into obstacle sensing, encoders into odometry and ros2_control, F/T and tactile into the manipulation controller, UWB and sonar into positioning.

7. Implementation

Turn the choice into a verified decision in five steps:

1. Write the requirement as numbers: range, accuracy, rate, environment, failure case to survive.
2. Shortlist with the Sensor Selection Calculator; read the cards’ “datasheet specifications to verify”.
3. Read the datasheet for each shortlisted model at your operating conditions (target reflectivity, temperature, mode, interface).
4. Check the interface, power and ROS 2 driver against the rest of the robot (the throughput and power guides).
5. Buy one, test it against the validation step in its entry, and only then standardise.

ROS 2 range, joint and wrench messages: verified facts

Message definitions read from the jazzy branch of ros2/common_interfaces on 2026-11-05. geometry_msgs/WrenchStamped was read from the same branch of the geometry_msgs package.

• sensor_msgs/Range: one reading from an active ranger (ultrasound or infrared radiation type), with field_of_view in radians (0 is the sensor x-axis), min_range, max_range and range in metres; values outside min and max are discarded. A fixed-distance (binary) ranger sets min_range = max_range and reports −Inf for a detection and +Inf for none (REP 117). It is not appropriate for laser scanners (use LaserScan).
• sensor_msgs/JointState: names with position (rad or m), velocity (rad/s or m/s) and effort (N·m or N); all arrays have the same size or are empty.
• geometry_msgs/WrenchStamped: a header and a wrench (force and torque); the header frame_id gives the reference frame, so the sensor frame must be defined in the ICD.
• There is no sensor_msgs type for radar, UWB or tactile arrays in this package; drivers use vendor or other packages’ messages, so check the driver’s message definition.

import math
def read_range(r, lo, hi, fixed=False):
    if fixed:
        return "detection" if r == -math.inf else "no detection" if r == math.inf else "invalid"
    return f"{r:.2f} m" if lo <= r <= hi else "discard (outside min/max)"
print(read_range(1.234, 0.02, 4.0))
print(read_range(5.0, 0.02, 4.0))
print(read_range(-math.inf, 0.3, 0.3, fixed=True))
print(read_range(math.inf, 0.3, 0.3, fixed=True))
js = {"name": ["j1", "j2"], "position": [0.1, 0.2], "velocity": [], "effort": [1.0]}
n = len(js["name"])
print({k: ("ok" if len(v) in (0, n) else f"size {len(v)} != {n}") for k, v in js.items() if k != "name"})

Printed (checked by the site’s tests):

1.23 m
discard (outside min/max)
detection
no detection
{'position': 'ok', 'velocity': 'ok', 'effort': 'size 1 != 2'}

Reading it: a reading outside min and max is dropped; a binary ranger returns −Inf for a detection and +Inf for none rather than a distance; and a JointState whose effort array has one element for two joints is malformed. Values are illustrative.

8. Real examples

9. Failure scenarios

• Radar (mmWave and automotive-class): Ghost objects from multipath.
• Ultrasonic range sensor: Crosstalk when several units fire together.
• Wheel / joint encoder: Wheel slip on loose or wet ground.
• Force/torque sensor: Overload damaging the sensor.
• Tactile sensor: Wear and delamination.
• UWB (ultra-wideband) ranging: Non-line-of-sight bias.
• Sonar (underwater acoustic): Multipath near the surface or bottom.

10. Troubleshooting

1. Encoders: drive a known distance and compare; check for missed counts near motor cables.
2. Ultrasonic: check crosstalk by firing one at a time.
3. Radar: place a corner reflector at known range and angle; check for ghost targets near metal.
4. F/T: re-zero and check drift with the tool at rest.

11. Validation

• Radar (mmWave and automotive-class): Place corner reflectors at known range and angle; check detection of a person and a static object; test next to metal.
• Ultrasonic range sensor: Measure flat, soft and angled targets at known distances; check crosstalk with all units firing.
• Wheel / joint encoder: Rotate the wheel a known number of turns and compare counts; drive a straight 10 m line and compare distance; check for missed counts under motor noise.
• Force/torque sensor: Zero and apply known weights; check drift over time and after temperature change; check overload protection.
• Tactile sensor: Press with known forces at known positions; check hysteresis and repeatability.
• UWB (ultra-wideband) ranging: Survey anchors; measure known points with line of sight and with an obstruction.
• Sonar (underwater acoustic): Measure a known target at known range in the water you work in; check the speed-of-sound setting.

12. Engineering checklist

• Regulatory approval for the radio sensors
• Clear field or acoustic path
• Encoder resolution and mounting side chosen
• Crosstalk plan for multiple ultrasonic units
• Overload protection and zeroing for F/T
• Anchor survey for UWB
• Driver and message type for vendor-specific sensors

13. Further references

• Class-level guidance written from general engineering knowledge; it contains no product specifications. Product-specific numbers must come from the manufacturer’s datasheet for the exact model and operating mode.
• ROS 2 message types: sensor_msgs (jazzy branch).
• ROS 2 message definitions (Jazzy): ros2/common_interfaces, sensor_msgs/msg, read 2026-11-05.

Related on FixTheRobot

• Complementary Strengths and Weaknesses of Cameras and LiDAR Perception · Lesson