Before introducing research robots
10 things to check
Rather than choosing a robot first, defining what to research and what results will be left can reduce overspecification and integration failure.
The key to introducing research robots is not choosing the most expensive equipment, but configuring a system that fits the research questions and verification procedures. If the base, sensor, controller, software, and safety equipment do not match each other, it is difficult to repeat the experiment even if the individual specifications are good.
1. Research objectives and success criteria
In one sentence, decide which is the key among autonomous driving algorithms, controller verification, sensor fusion, multi-robot operation, and manipulation research. Indicators to be measured, such as target speed, route error, and completion time, as well as allowable failure conditions are also written down.
2. Real-world environment and movement method
Record indoor/outdoor, floor material, slope, threshold, aisle width and turning space. Differential drive, omnidirectional, and independent steering have different control issues and space requirements, so comparisons should not be made solely on external size.
3. Payload and center of gravity
Calculate the total load including not only sensors and PC, but also arms, batteries, cables, and protective structures. When lifting a moving arm or a long object, the change in center of gravity and the risk of tipping become more important than the static load.
4. Sensor and time synchronization
Check the mounting location, data cycle, and timestamp standards of LiDAR, cameras, IMU, and encoders. We separate sensor quality and calibration procedures to avoid mistaking the position estimation problem as an algorithmic problem.
5. Control interface
Check the scope of CAN, Ethernet, Serial, ROS 2 drivers, SDK and low-level control access. If the layer the researcher needs to change is locked by the manufacturer, the desired experiment may not be possible.
6. Power and operating time
Check battery voltage, continuous operation time, charging time, and how to power on and off external equipment. Power shortages and thermal issues significantly reduce repeatability in long experiments.
7. Safety stop system
We need to test what state E-stop makes the base, arm, power and control PC. Document speed limits, safety scanners, protective stops and recovery procedures.
8. Simulations and models
Check out our range of CAD, URDF, mass/inertial, and sensor models and simulation packages. Differences between the actual system and the model must be recorded in order to move from SIL to RIL/actual vehicle.
9. Maintenance and reproducibility
Find spare parts, firmware versions, driver support, log collection and settings backup methods. The value of a research asset lies in whether it can be successful again under the same conditions rather than a demonstration that was successful once.
10. Final result
Decide before signing the contract whether you will receive only the equipment or whether you will need configuration diagrams, BOM, sources, experiment scenarios, logs, training, and results reports. The scope of exclusion must also be small so that schedules and costs can be judged realistically.
next steps
RODIX first confirms the research goals and environment and proposes platform configuration, safety conditions, and 90-day verification scope. The next steps are judged based on measurable results without guaranteeing completed performance in advance.
Note
We referred to RODIX's internal mobile manipulator research report, future automobile industry-academia project presentation materials, and ROS 2·Nav2·MoveIt 2 public documents.