Le 10 domande tecniche più frequenti che vi ponete sui robot mobili (speciale Ricerca & Sviluppo e Università)

10 frequently asked technical questions about mobile robots (specialising in R&D and higher education)

  • 8 April 2025

For several years, Génération Robots has been working with research laboratories and universities on their robotics projects. We provide development platforms, mobile robots and solutions based on ROS / ROS 2. Projects often evolve: after a testing phase, they move on to concrete applications that pose real technical challenges.

 

Here are 10 advanced questions that our customers regularly ask, with answers from our engineers.

1) Can the software architecture of a mobile robot be adapted to a custom framework (beyond ROS/ROS 2)?

Yes, provided you choose an open platform. Most professional mobile robots (AgileX, Husarion, Clearpath, etc.) are based on ROS or ROS 2, which allows for fairly extensive customisation.To go further, some platforms offer low-level access via CAN, UART, or REST/WebSocket APIs, allowing you to rebuild a completely custom software architecture.

2) How can sensor fusion (IMU, LiDAR, odometry) be optimised for more stable navigation in dynamic environments?

ROS packages provide a good foundation, but for dynamic environments, it is essential to:

  • Fine-tune sensor covariances
  • Add visual perception (visual SLAM or RGB-D)
  • Integrate an adaptive module (through learning or heuristics) to adjust fusion according to context

Precise temporal synchronisation between sensors is also essential.

3) Is it possible to perform multi-robot SLAM in real time?

Yes, but this remains a complex issue. Tools can be adapted for multi-robot use. To do this, you need to:

  • Synchronise clocks (NTP, GPS)
  • Manage map sharing (map merging or cloud map)
  • Use ROS 2 to leverage DDS and namespaces

A distributed and well-isolated architecture is essential to ensure system stability.

4) What are the current limitations of AMRs for unstructured environments (outdoor, semi-natural)?

The main limitations are:

  • The lack of reliable landmarks for SLAM
  • The difficulty of modelling uneven terrain
  • Resistance to weather conditions (rain, dust, bright light)
Platforms such as the Husky, Ranger Mini 3.0 and Panther are suited to these constraints, but must be accompanied by robust perception and planning algorithms.

5) Which mobile robot do you recommend for advanced development in multi-agent robotics?

Several platforms are suitable for these uses:

They all allow for preliminary simulation (Gazebo, Webots) and good ROS 2 integration.

White Paper Mobile Manipulator Robots

6) What strategies are there for autonomous recharging in the laboratory (without heavy infrastructure)?

Here are some practical strategies:

  • DIY load station with visual alignment or using April Tag
  • 3D-printed conductive connectors (pogo pins)
  • Inductive charging if the platform allows it

Everything must be integrated into the robot’s state machine, with autonomous decision-making based on battery level and station availability.

7) What are the best practices for logging and replaying complex missions?

The right tools:

  • rosbag or ros2 bag for topic recording
  • PlotJuggler for real-time or delayed visualisation
  • InfluxDB + Grafana for centralised fleet logs
  • RViz or Foxglove Studio for replaying trajectories

Don’t forget to synchronise the clocks on each robot (NTP, GPS) to ensure data consistency.

8) What tools do you recommend for simulating fleets or autonomous behaviours?

  • Gazebo/Ignition: advanced physical simulation
  • Webots: lightweight, multi-agent, very well documented
  • Isaac Sim (NVIDIA): excellent for embedded AI and deep RL
  • Stage/RViz2: useful for quick navigation tests

It all depends on the level of realism and the need for visual or physical rendering.

9) Are robotic platforms easily expandable with robotic arms?

Yes, provided that you:

  • Respect the payload and centre of gravity
  • Adapt the mounting base (VESA, rail, custom plate)
  • Couple the planners (MoveIt + Nav2)
  • Synchronise coordinates between base and arm
Some platforms, such as Robotnik’s RB-KAIROS+, are specially designed for this purpose.
Mobile robot buying guide

10) What is the level of cybersecurity for ROS/ROS 2 mobile robots?

ROS 1 does not natively integrate any security mechanisms. ROS 2, on the other hand, relies on DDS to offer to:

  • Respect the payload and centre of gravity
  • Adapt the mounting base (VESA, rail, custom plate)
  • Couple the planners (MoveIt + Nav2)
  • Synchronise coordinates between base and arm

We also recommend:

  • Isolation of the robot network (VLAN, firewall)
  • Regular access audits
  • The use of certificates for sensitive communications

Conclusion

Raising these issues at the design stage of a project saves time and avoids numerous technical bottlenecks. At GR Lab, our engineering design office, we regularly work on these types of issues to help researchers and engineers move forward more quickly. What challenges are you currently facing in your robotics projects?
Marine Senecat, Content Manager chez Génération Robots

Marine Senecat

Content Manager at Generation Robots