PROJECT / 02 · Controls · Sensing · Experimental robotics

Control for a compact inspection manipulator

A riser-deployed cable-driven articulated manipulator for confined inspection, with external joint sensing, simulation and motion planning, and experiments on arm-joint control.

FIU Applied Research Center · Collaboration with INL and DOE project teamsResearch and physical experiments

Python
C++
Joint encoders
PI control
ROS 2
Isaac Sim

Simulated riser-deployed inspection in a representative pit, alongside camera, depth, and point-cloud views. Recorded in Isaac Sim.

01 / THE CHALLENGE

A physical engineering problem

A cable-driven mechanism can receive a motor command without immediate joint motion. Breakaway depends on the joint, direction, command, and recent motion history. Joint-level sensing is needed to distinguish a command from a physical response.

02 / MY CONTRIBUTION

What I worked on

Develop kinematic models and workspace analysis for an articulated manipulator designed for inspection through 4-inch risers.
Integrate external joint encoders with Python/C++ control and experimental software.
Build ROS 2, MoveIt 2, and Isaac Sim workflows with robot control interfaces, a representative pit model, and collision-aware planning.
Design and analyze breakaway experiments on arm joints J1–J4 and develop bounded assistance that releases when encoder measurements confirm motion.

03 / SYSTEM WORKFLOW

Breakaway assistance at the arm joints

01
Measure
External encoders observe the joints
02
Characterize
Command, direction, and motion history
03
Assist
Bounded support alongside nominal PI control
04
Confirm
Release on measured motion or timeout

04 / ENGINEERING DECISIONS

Decisions behind the implementation

Connect the simulated robot to a planning scene

The control framework connects Isaac Sim and ros2_control with ROS 2 topics and MoveIt 2/OMPL. Pit geometry supplies collision context for planning, while simulated camera views and point clouds support inspection development.

Measure at the output

External joint encoders report the mechanism's actual response. Motor commands alone do not establish that a joint has started moving.

Use experiments to inform the intervention

The experimental program includes 400 duty-level trials and 576 motion-history trials on the four arm joints, J1–J4. The full inspection platform also includes base and tool joints.

Keep assistance local and bounded

The control approach supplements nominal PI control during persistent sticking, with encoder-confirmed release and timeout handling.

PROJECT GALLERY

Laboratory hardware and simulation

Laboratory platform with original annotations identifying the base, arm joints J1–J4, tool joint, actuators, and control box.

Hands-on integration of the articulated inspection robot with the laboratory team.

Control framework: Isaac Sim, ros2_control, ROS 2 interfaces, MoveIt 2/OMPL, and the pit collision scene. Original figure from the presentation, slide 8.

The articulated robot in the representative pit model in Isaac Sim. Presentation, slide 8.

Corresponding MoveIt planning scene with the modeled pit geometry. Presentation, slide 8.

Kinematic and collision-free end-effector reachability density in the modeled pit. These are simulation results. Original figure from the presentation, slide 13.

05 / OUTCOME

What the work demonstrates

Analyzed 976 breakaway trials and developed a control approach grounded in measured joint behavior. My focus spans the sensing and control software, robot modeling, experimental design, and validation.

Collaborative robotics research at FIU ARC with Idaho National Laboratory and DOE project teams. Presentation figures are from “Simulation-Driven Control and Workspace Analysis for a Riser-Deployed Articulated Robot in Hanford Waste Tank Pits,” with FIU and INL co-authors. The inspection video and workspace figures show simulation of a representative pit.

EXPLORE ANOTHER PROJECT

Classroom robotics platform

Get in touch