Designing the hardware behind mobile manipulation
Ongoing CERLAB work on a shape-changing end effector for smoothing 3D-printed concrete, combining mechanical design with control and path planning.
To make a concrete-printing tool adapt its surface shape to the material and toolpath instead of relying on a fixed smoothing profile.
From the initial question to a tested result
The timeline traces how the project moved through problem definition, design, prototyping, integration, and validation.
Treat the robot as one integrated physical system.
My role sits at the robotics–mechanical-design boundary. I am contributing to the physical platform and its integration rather than treating the manipulator, mobile base, environment, and controls as isolated pieces.
Design for both robot performance and research iteration.
Research hardware needs to be changed, instrumented, and tested repeatedly. The mechanical work therefore emphasizes accessible interfaces, modular CAD, realistic fabrication constraints, and components that can be revised without rebuilding the entire platform.
A mechanism is useful only when it improves the full manipulation task.
As the work progresses, this chapter will connect design choices to integration tests: fit, range of motion, interference, stability, repeatability, and performance during representative manipulation tasks.
Three decisions that shaped the project
Each decision connects a technical constraint to the choice I made, the analysis behind it, and the tradeoff that followed.
Surface shaping
Five linear actuators
A fixed tool cannot follow the changing geometry of printed concrete. Five independently driven points provide enough control to reshape the surface while keeping the mechanism understandable.
Added controls complexity to create an actively variable smoothing profile.
Treat the robot as one integrated physical system.
My role sits at the robotics–mechanical-design boundary. I am contributing to the physical platform and its integration rather than treating the manipulator, mobile base, environment, and controls as isolated pieces.
- Translate task and workspace needs into mechanical requirements.
- Account for interfaces between structure, actuation, electronics, sensing, and controls.
- Keep design decisions traceable as the research platform evolves.
Tool face
Flexible metal surface
A continuous metal face avoids discrete actuator tips marking the concrete and converts point motion into a smooth working profile.
Enabled continuous curvature while requiring careful stiffness and force calibration.
Treat the robot as one integrated physical system.
My role sits at the robotics–mechanical-design boundary. I am contributing to the physical platform and its integration rather than treating the manipulator, mobile base, environment, and controls as isolated pieces.
- Translate task and workspace needs into mechanical requirements.
- Account for interfaces between structure, actuation, electronics, sensing, and controls.
- Keep design decisions traceable as the research platform evolves.
Development path
Model before integration
The coupled risks in actuator motion, surface deformation, and toolpath planning were easier to isolate in a model before adding manipulator kinematics and hardware constraints.
Delayed manipulator mounting to reduce risk in geometry, motion, and path planning first.
Design for both robot performance and research iteration.
Research hardware needs to be changed, instrumented, and tested repeatedly. The mechanical work therefore emphasizes accessible interfaces, modular CAD, realistic fabrication constraints, and components that can be revised without rebuilding the entire platform.