Differential Suspension

Differential Suspension – definition

In mobile robotics, a differential suspension is a passive mechanical linkage that allows the left and right sides of a wheeled chassis to redistribute vertical load when one wheel encounters an obstacle or a depression. It is not the same as differential drive. Differential drive describes a kinematic propulsion layout controlled by wheel speed difference. Differential suspension describes how wheel contact is mechanically equalized across uneven terrain.

For a UGV, the practical goal is simple: keep more wheels in ground contact, reduce peak pitch and roll of the body, and improve traction without requiring active suspension actuators. In small and medium off-road robots, differential suspension is often implemented as a rocker, rocker-bogie, or linked axle mechanism with a central pivot or differential joint. When one side of the suspension rotates upward, the linkage transfers part of that motion to the opposite side. This reduces body attitude change compared with a rigid chassis.

In the context of Leo Rover and Raph Rover, the term matters because terrain mobility directly affects odometry quality, slip, sensor stability, and navigation reliability. A UGV running ROS 2 can estimate pose from wheel encoders, IMU, LiDAR, visual odometry, or GNSS. All of these sources can degrade when the chassis loses contact or oscillates excessively. Differential suspension does not solve localization on its own, but it improves the mechanical conditions under which localization and control operate.

How differential suspension works on a UGV

The mechanism is passive. No motor command is required. A typical layout has wheels mounted on arms or bogies that can rotate relative to the chassis. The left and right suspension members are mechanically coupled by a differential pivot, equalizer bar, or central rocker link. If the front-left wheel climbs a rock, the left suspension side rotates. Instead of transferring the full disturbance directly to the chassis, part of the motion is compensated by the coupled right side.

This behavior changes the vehicle response in three important ways.

It increases contact retention on rough ground.
It reduces sudden body angle changes seen by IMU and cameras.
It lowers the probability of one wheel unloading completely during obstacle crossing.

The traction effect can be summarized by the basic friction relation:

F_traction ≤ μ · N

Here, μ is the tire-ground friction coefficient and N is the normal force on the wheel. A differential suspension does not increase μ, but it can help distribute N more evenly across wheels. On deformable soil, gravel, ramps, and field terrain, that often improves usable traction and reduces wheel slip.

Differential suspension versus differential drive

These terms are frequently confused because both use the word “differential”. In robotics documentation, they refer to different subsystems. This distinction is important when reading URDF models, controller configuration, and hardware specifications.

Term	Subsystem	Main function	ROS 2 relevance
Differential drive	Locomotion / kinematics	Steer by left-right wheel speed difference	`diff_drive_controller`, `geometry_msgs/msg/Twist`, odometry
Differential suspension	Chassis mechanics / mobility	Equalize wheel loads and improve terrain following	Indirect effect on odometry, IMU data, perception stability

Leo Rover uses a differential drive propulsion concept. Whether a specific rover chassis also has a differential suspension depends on its mechanical design, not on the drive controller in ROS 2.

Impact on ROS 2 localization, odometry, and perception

ROS 2 does not define a standard message called “differential suspension”. The effect appears indirectly through state estimation and sensor data quality. The most relevant interfaces are odometry, IMU, joint states, transforms, and optionally contact or suspension joint telemetry if instrumented.

On a terrain UGV, body motion couples into several ROS 2 data streams:

nav_msgs/msg/Odometry – wheel odometry becomes less reliable when slip rises during wheel unloading.
sensor_msgs/msg/Imu – pitch and roll disturbances increase linear acceleration and angular velocity noise.
sensor_msgs/msg/LaserScan or sensor_msgs/msg/PointCloud2 – LiDAR motion distortion and scan misalignment can increase if the platform oscillates.
tf2 transforms – frame stability assumptions are harder to maintain if the sensor mast moves relative to the base.

In ROS 2 Humble and newer, a typical estimation stack may combine encoder odometry with IMU using robot_localization. A passive suspension that preserves contact and reduces shocks helps both sources. This is especially relevant on Leo Rover, where compute resources are often limited by a Raspberry Pi-class onboard computer, so robust mechanics often matter as much as algorithmic complexity.

ekf_filter_node:
  ros__parameters:
    frequency: 30.0
    two_d_mode: true
    publish_tf: true
    odom0: /wheel/odometry
    odom0_config: [true, true, false,
                   false, false, true,
                   true, true, false,
                   false, false, true,
                   false, false, false]
    imu0: /imu/data
    imu0_config: [false, false, false,
                  false, false, true,
                  false, false, false,
                  false, false, true,
                  true, false, false]

This configuration does not model suspension dynamics explicitly. It benefits from them mechanically.

Key parameters and engineering metrics

When evaluating a differential suspension for a research UGV, the relevant metrics are mechanical first and algorithmic second. Manufacturers may not publish all of them, so some are measured experimentally during field tests.

Wheel travel – vertical displacement range of each wheel relative to the chassis.
Articulation angle – maximum rotation of rocker or bogie members.
Ground clearance – minimum clearance under static and articulated conditions.
Static stability margin – resistance to tip-over on side slope or obstacle edge.
Load distribution – how normal force is transferred among wheels during articulation.
Obstacle surmount capability – often given as a maximum step or rock height.
Body attitude disturbance – pitch and roll change while crossing a defined obstacle.

For robotics experiments, two measurable integration-level indicators are often more useful than a pure mechanical drawing:

Wheel slip ratio during obstacle crossing, derived from encoder speed versus estimated body velocity.
Localization drift per meter on uneven terrain, measured from fused ROS 2 odometry against ground truth.

Practical context for Leo Rover and Raph Rover

Leo Rover is a compact 4-wheel differential drive platform intended for research and education. It is suitable for testing perception, teleoperation, ROS 2 nodes, navigation pipelines, and outdoor sensing, but autonomous field performance still depends on the full integration stack. If a payload includes a depth camera, 2D LiDAR, or RTK GNSS, stable wheel contact and reduced body oscillation directly improve data consistency.

On Leo Rover, a differential suspension should be considered in relation to payload mass, center of gravity, and available compute. A heavy mast-mounted sensor can increase roll moment and amplify terrain disturbance. In such a case, passive articulation can be more valuable than adding algorithmic filtering alone.

Raph Rover addresses use cases with higher payload and harsher terrain. In this class of platform, suspension design has stronger consequences for traversability, frame rigidity under load, and sensor survivability. A differential suspension can support larger LiDAR units, GNSS antennas, and inspection payloads by reducing impact transfer to the main body. It does not replace careful frame design, wheel selection, or state estimation tuning.

Modeling and simulation considerations

In URDF, a differential suspension is represented by links and joints, not by a drive plugin. For simulation in Gazebo or other physics engines, correct joint limits, inertia tensors, damping, and contact parameters are required. A visually plausible model is not enough. Poor inertial modeling will produce unrealistic oscillations and wrong contact forces.

<joint name="rocker_left_joint" type="revolute">
  <parent link="base_link"/>
  <child link="rocker_left"/>
  <origin xyz="0 0.12 0" rpy="0 0 0"/>
  <axis xyz="1 0 0"/>
  <limit lower="-0.45" upper="0.45" effort="30" velocity="1.0"/>
  <dynamics damping="0.2" friction="0.05"/>
</joint>

If the suspension is uninstrumented, ROS 2 may expose only wheel joints and IMU data. If articulation sensing is available, joint positions can be published through sensor_msgs/msg/JointState. That can be useful for terrain classification, chassis state monitoring, or adaptive speed control.

Limitations and trade-offs

Differential suspension improves mobility, but it introduces design trade-offs. More joints mean more mechanical complexity, more wear points, and tighter requirements for sealing and maintenance. For IP-rated outdoor robots, the linkage must tolerate mud, dust, and water ingress. It can also complicate precise sensor extrinsic calibration if payload supports flex relative to the main frame.

There is also a systems trade-off. Better terrain following can reduce shocks, but a very compliant structure may increase low-frequency body motion. That can affect camera-based visual odometry and horizon stabilization. The best design depends on terrain profile, payload mass, wheel diameter, and target speed.

Normative references and standards

There is no single ROS REP dedicated specifically to differential suspension. The relevant standards and specifications are distributed across robot description, frame conventions, and message interfaces used to describe the chassis and estimate its state.

REP 103 – Standard Units of Measure and Coordinate Conventions, ROS.org.
REP 105 – Coordinate Frames for Mobile Platforms, ROS.org.
URDF – Unified Robot Description Format, ROS documentation.
sensor_msgs/msg/Imu, sensor_msgs/msg/JointState, nav_msgs/msg/Odometry – official ROS message definitions.
ros2_control and diff_drive_controller – relevant for propulsion control, not for suspension itself.

For mobile robot research, IEEE literature on rocker-bogie and articulated suspension systems is the most appropriate source for obstacle climbing, stability, and traction analysis. Manufacturer specifications should be used for wheel diameter, payload, chassis mass, and environmental rating of the target platform.