On Shaky Ground: Characterising Non-Rigid Base Effects in Mobile Manipulation
Authors: Zongyuan Zhang, Christopher Lehnert, Will N. Browne, Jonathan M. Roberts
Organizations: Faculty of Engineering, Queensland University of Technology, 2 George St, Brisbane, QLD 4000, Australia · Australian Cobotics Centre, Queensland University of Technology, 2 George St, Brisbane, QLD 4000, Australia · QUT Centre for Robotics, Queensland University of Technology, 2 George St, Brisbane, QLD 4000, Australia
Mobile robot manipulators extend contact-task capability beyond the workspace of fixed robot installations, but their bases must also sustain the forces generated by the task. On lightweight platforms, base motion and support compliance can change the force and motion realised at the workpiece. How these non-rigid base effects manifest during sustained contact remains insufficiently characterised. This paper investigates non-rigid base effects in mobile robot manipulators through a sanding case study combining exploratory simulation with experiments on two lightweight, unactuated wheeled testbeds. Joint measurements of contact force, tool motion, and base vibration show that sustained contact coexisted with substantial task error: under low base-ground resistance, only 54-70% of a 100 N command was realised, with tool-centre-point errors reaching 100 mm. Increasing base-ground resistance during contact improved force realisation and arrested further tool displacement, while base vibration did not vary consistently with force-tracking error in the tested conditions. These observations illustrate how non-rigid base behaviour can affect contact-task execution and highlight the value of examining force, motion, and vibration together when characterising lightweight mobile manipulation.
Figures & tables
Figure 1: Contact forces acting on a mobile manipulator during sanding. The contact forces Fn and Ft and the tool torque τtool are transmitted through the manipulator to the base and balanced by the ground support forces N1 – N3 and the friction forces Ff , with the weight mg acting at the centre of mass (CoM). Limited traction and compliance at the base–ground interface can lead to base motion, loss of stability, and vibration.
Figure 2: Simulation environment and representative non-rigid base effects.
Simulation
Cart
Trolley
Arm
UR10 model
UR10
UR5
Base
Rigid wheels
Aluminium
Plywood
Support
Rigid floor
Concrete
Foam
Resistance
Varied
Low, high
Low, high
Sander speed
Varied
Low, high
High
Table 1: Simulation model and physical platforms.
Figure 3: Physical testbeds. The cart has an aluminium-extrusion frame with a slotted aluminium deck; the trolley has a plywood deck with the compliant layer beneath the base. Panel (c) shows the force/torque sensor, deck-mounted inertial measurement unit, tripod-mounted camera, and tool-mounted ArUco marker used with the trolley; the same sensors were deployed on the cart.
Condition
Result
ID
Resistance
Posture
Speed
F∗
Fˉ
ρF
eF
σF
Slope
eTCP
app (m s -2 )
(N)
(N)
(N)
(N)
(N s -1 )
(mm)
med.
max
Experiment 1 (cart)
T1
Low
Standard
Low
100
58.8
0.59
41.5
2.6
0.42
n/a
0.19
0.69
T2
Low
Low
Low
100
68.3
0.68
32.1
3.7
0.32
n/a
0.08
0.12
T3
High
Standard
Low
100
94.3
0.94
6.9
3.2
0.21
n/a
0.29
0.77
Table 2: Trials by experiment, with force realisation, tool centre point (TCP) error, and base vibration during contact, using the metrics defined in Section 4.5 .
Figure 4: Experiment 1: single-condition trials, each shown during contact up to retraction. Top: measured normal force and command. Bottom: base acceleration envelope normal to the deck. In each row, (a)–(d) share the vertical axis.
Figure 5: Experiment 2: switch from low to high resistance part-way through contact. The bar above each panel marks the low-resistance (shaded) and high-resistance phases; the dotted line marks the switch. Top: normal force. Middle: TCP error, the displacement of the tool marker from its position at the start of the record, tracked by the fixed camera. Bottom: base acceleration envelope normal to the deck. In each row, (a) and (b) share the vertical axis.
Figure 6: Experiment 3. Top: normal force. Middle: TCP error, the displacement of the tool marker from its position at the start of the record. Bottom: base acceleration envelope normal to the deck. In each row, (a) and (b) share the vertical axis.
Aspect
Simulation
Hardware observations
Base motion
Slip when the required wheel–ground tangential forces exceed the available traction
About 100 mm TCP error at low resistance (T5–T7), accompanied by ρF of 0.54–0.70 (T1, T2, T5–T7), on both platforms
Posture sensitivity
Extended posture increases excitation and reduces the stability margin
Low posture reduced median base vibration from 0.19 to 0.08 m s -2 and raised ρF from 0.59 to 0.68 (T1, T2)
Base–ground interaction
Higher wheel friction suppresses slip
High resistance raised ρF to 0.88–1.01 on both platforms (T3, T5, T6, T8); the switch arrested TCP-error growth within about 1 s (T5, T6)
Oscillation / tool excitation
Tool excitation produced tool and base oscillation; tool speed shifted the dominant frequency
High tool speed produced substantially higher base vibration than low tool speed, both out of contact and at high resistance (T5, T6)
Support compliance
Not represented in the rigid-wheel model
Foam nearly tripled median base vibration relative to the hard floor (Figure 7 )
Loss of stability
Occurred when the zero-moment point moved outside the support polygon
Not reached
Table 3: Qualitative comparison of simulated behaviours and hardware observations.
Figure 7: Experiment 3, support comparison (low resistance, 20 N command): base acceleration envelope normal to the deck.
Contact-rich manipulation requires robots to balance accurate motion tracking with compliant interaction, yet most visual-action policies leave compliance fixed at the controller level. We present Imp-ACT, a methodologically grounded and practical approach to incorporating direction-dependent Cartesian stiffness modulation directly into demonstration collection, without manual stiffness selection or offline target reconstruction. During teleoperation, a self-tuning impedance controller adapts stiffness along the instantaneous direction of motion while maintaining compliance in orthogonal directions. The adapted stiffness is applied and recorded alongside visual observations and motion commands, capturing motion and compliance under the same dynamics. We implement this pipeline using Action Chunking with Transformer (ACT) to predict end-effector pose, gripper action, and motion-direction stiffness from visual, proprioceptive, and wrench observations. The performance of Imp-ACT is evaluated on wiping and plug insertion using both success rate and quantitative measures of contact behavior. Compared with fixed low- and high-stiffness baselines, Imp-ACT achieves comparable or higher success while maintaining low interaction forces. In wiping, it reduces contact-force vibration by approximately 29× relative to the compliant baseline and 180× relative to the stiff baseline. In plug insertion, it reduces forces orthogonal to the insertion direction by 43% relative to the better fixed-stiffness baseline. These results highlight the benefit of maintaining sufficient stiffness along the direction needed for task execution while preserving compliance in other directions to limit contact forces and accommodate environmental constraints.
Luca Zanetti, Doganay Sirintuna, Idil Ozdamar +3
Human-Robot Interfaces and Interaction Laboratory, Istituto Italiano di Tecnologia, Genoa, Italy · Safe AI Lab, Carnegie Mellon University
Contact-rich manipulation underlies many everyday human activities, yet evaluating robots on such tasks remains difficult: the tasks involve diverse contact behaviors and are hard to reproduce across laboratories. Drawing on observations of everyday human manipulation, we identify eight essential skills that provide a practical framework for studying contact-rich manipulation. We introduce ContactBench 1.0, an extensible benchmark that instantiates these skills through 36 reference tasks and more than 40 object assets. The assets capture task-relevant geometry and contact behavior, can be fabricated reproducibly on consumer-grade 3D printers, and are each paired with a matching simulation model. Controlled task variations in both the real world and simulation enable evaluation of how well robot systems generalize. We evaluate recent representative vision-only and visual-tactile policy learning methods on six benchmark tasks. The results reveal that while vision-only policies struggle to perceive or react to contact events, visual-tactile policies do not meaningfully improve their performance, highlighting that contact-rich manipulation remains a significant challenge for current policy learning methods. By providing common task definitions alongside physical and simulated assets, ContactBench 1.0 offers a shared testbed for developing, evaluating, and comparing contact-rich manipulation systems. The benchmark will be open-sourced at https://jingyi-xiang.github.io/contactbench.
Jingyi Xiang, Yixuan Wang, Jeannette Bohg +1
University of Illinois Urbana-Champaign, Urbana, IL, USA · Stanford University, Palo Alto, CA, USA
Teleoperation is widely used to collect high-quality demonstrations for robot learning. Dexterous manipulation, however, requires coordinated control of motion and contact forces. Operators can specify motion, but struggle to reproduce the subtle force adjustments of human manipulation without feeling the robot's contacts. We introduce Force-Modulated Dexterity (FOMO-Dex), a unified compliance formulation for contact-rich manipulation. FOMO-Dex expresses nominal actions in coordinates defined by task geometry and contact structure. It applies force-driven motion corrections along selected directions while leaving the remaining command components unchanged. Across card picking, paper cutting, screwdriving, and nut-and-bolt assembly, FOMO-Dex reduces excessive contact forces compared with stiff control. A user study shows that operators collect more successful demonstration data per hour after brief practice. For policy learning, imitating the corrected actions improves card-picking success over imitating raw operator commands. Policies trained on FOMO-Dex demonstrations also achieve more rotation in nut-and-bolt manipulation and screwdriving than those trained on demonstrations collected with the stiff controller. Videos are shown on https://yingyuan0414.github.io/fomo-dex/ .