8. Conclusion

This article focused on methodologies for dynamic identification of industrial robots. Today, for conventional robots with rigid structures (industrial serial manipulators, gantries, SCARA robots), the identification of dynamic parameters can be considered a solved problem: robust and proven methods make it possible to build reliable and accurate models for most industrial applications.

However, many challenges remain for less conventional robots such as collaborative robots, exoskeletons, and cable robots. Their often lightweight and economical design, combined with their structural flexibility, introduces complex nonlinear behaviors (bending, mechanical play, wear) that make identification more difficult and estimated parameters less accurate.

At the same time, recent advances in reinforcement learning (RL) are generating considerable...