**7. Conclusions**

The here presented paper describes the mechanical and control design solutions for (i) a low-cost hardware industrial exoskeleton (ii) with high payload ratio to be adopted in lifting and transportation of heavy parts. Mechanical design specifications have been derived from the task, allowing to design (iii) an intrinsic compliant 2 DoFs exoskeleton exploiting e SEA actuation at the shoulder joint to intrinsically increase human-robot interaction safety. The proposed control architecture has been described, defining (iv) a safety-based control framework. The inner gain scheduling optimal controller allows for task trajectory tracking. The outer safety-based fuzzy logic controller allows for human empowering. Simulation results show promising performance in the assistance of human operators (damping vibrations and empowering workers) and in the manipulation of unknown payloads.

The prototype of the proposed solution is under realization and it will be experimentally tested in the proposed task to evaluate the proposed approach. In particular, 20 subjects will be considered in the experimental tests. Both cognitive evaluation (based on questionnaires) and quantitative evaluation (based on EMG measurements) will be performed as in Reference [49]. In addition, authors will apply to the second call of the *EUROBENCH project* to test the proposed exoskeleton in the EUROBENCH exoskeletons benchmark facility, where the leading author Roveda is also leading the *STEPbySTEP project* [51].

#### **8. Current and Future Work**

Considering the mechanical design of the device, current work is focusing on a 3 DoFs shoulder joint concept, implementing 2 additional passive DoFs. In this new concept, the shoulder motor is considered aligned with the shoulder joint. The resulting exoskeleton implements, therefore 4, DoFs (Figure 14). The main advantage of the proposed new design is related to the increased mobility of the shoulder. However, such joint requires a different design of the compliant actuation. Therefore, the here mentioned solution is still under evaluation. Additional work is devoted to design a passive ergonomic back support for the exoskeleton to increase the ergonomics of the device. Considering the control design of the device, machine learning techniques are investigated to optimize the outer controller parameters.

**Figure 14.** Three DoFs shoulder joint: new concept to increase exoskeleton mobility and task DoFs.

**Author Contributions:** Conceptualization, L.R., J.L., G.F., D.F., M.M. and A.M.; methodology, L.R.; software, A.M., J.L. and L.R.; validation, A.M., J.L. and L.R.; formal analysis, A.M., J.L. and L.R.; investigation, J.L. and L.R.; resources, L.R.; data curation, L.R.; writing–original draft preparation, J.L., A.M. and L.R.; writing–review and editing, L.R.; visualization, L.R.; supervision, M.M., F.B. and G.L.; project administration, L.R.; funding acquisition, L.R.

**Funding:** The work has been developed within the EFFORTLESS project, funded from CNR-STIIMA. This project has received funding from the European Union's Horizon 2020 research and innovation programme, via an Open Call issued and executed under Project EUROBENCH (grant agreement No 779963).

**Acknowledgments:** Authors would like to thank Tito Dinon (CNR-STIIMA) for his expertise and support in the project.

**Conflicts of Interest:** The authors declare no conflict of interest.
