Collect dataset for action recognition for the lifting task

[kouroshD]

In this issue, following the discussion for the risk assessment, we want to collect data for human lifting an object.

[kouroshD]

@lrapetti can you please, drop here the link to the dataset and the video of the dataset we collected. Thanks.

[kouroshD]

We did the experiments and collect data for risk assessment purposes, on September 30th. For this purpose, the user should lift a 5kg object from the low height (20cm above the floor level) and put it a table (with apporximately 1.3 m height, the lab front desk where we put the different team tools). @lrapetti Was standing, bending to grasp the object, graping the object, lifting it, putting it on the table, going to stand pose, taking the object again, and putting it down on the box 20cm above the floor.

The bending has been done in two ways:

• Ergonomic way: The human was bending his knee to grasp the object and leave it.
• Non-ergonomic way: The human was bending with the back (mainly hip and back spine), and knee was straight in order to bend and grasp the object.

The data have been collected continuously for the two approaches, each for 5 mins approximately, and in total 10 mins.

[lrapetti]

Video and data can be found in https://istitutoitalianotecnologia.sharepoint.com/:f:/r/sites/ergoCub/Documenti%20condivisi/General/experimental-datasets/2021-09-30-human-lifting-data?csf=1&web=1&e=bPgxTp

I have also taken some geometric measurements after the experiment

[kouroshD]

We did this, so we can close this issue.