4. Conclusion

We presented a range of methods for programming a robotic device to assist with walking and standing up.

The methods described were tested in real-life conditions with people who had difficulties getting up and/or walking. We were thus able to identify the importance of movement prediction in the control law for walking. Similarly, we saw that intention detection combining the interaction forces on the handles and the orientation of the trunk in the sagittal plane was the most functional for assisting with standing up. Methods for making measurements more robust, such as Kalman filters or data fusion, are avenues for achieving a reliable robotic assistance system.