Requirements:
Data files are in .xlsx format
Scripts for plots require matlab2020 +
Scripts for Injury risk metrics require matlab2020 +
Setup:
git clone https://github.com/epfl-lasa/service_robots_collisions.git
Dowload the Dataset with raw and processed sensor information from: https://doi.org/10.5281/zenodo.5266447 [1]
Create a folder on the parent directory: "../collision_data"
Place the following folders from the dataset in this directory:
collision_test_analysis
collision_test_data
data_metrics
data_raw
This repository includes data of collisions between a service robot - Qolo - and pedestrian dummies: male adult Hybrid-III (H3) and child model 3-years-old (Q3). You will find scripts to read and plot the data, as well as, analysis of the injury risk based on standard crash testing metrics: Head Injury Criteria (HIC-15), Neck Injury (Nij), Chest deflection, and tibia forces.
Further descriptions of the scenarios included in this data can be found in the supplementary materials of the referenced publication.
All data should be contained on the parent directory on a folder with the name "collision_data", for direct execution of the functions provided. Alternatively, give the directory location of the dataset downlowded from [2].
structure: Test#/01_values/test_num_name_CFC1000 (xlsx) --> This files contain the data processed with CFC1000 filter (considered Raw for the current sensor setup)
Raw data in a single struct for Adult 50-percentile Dummy H3: data_raw/H3_raw_collision_struct.mat
Raw data in a single struct for Child Dummy Q3: data_raw/Q3_raw_collision_struct.mat
Filtered data: data_raw/filtered_collision_struct.mat
% List of data recordings
% % Each scenario comprises 3 speeds at contact --> [1.0, 1.5, 3.1] [m/s]
% 4 setups with the child dumm Q3 (3-years-old dummy) [1.05m height / 15kg weight]
% % Setup A: Dummy Q3 impact at the Chest - [133kg carrier robot]
% % Setup A2: Dummy Q3 impact at the Chest - [60kg carrier robot]
% % Setup B: Dummy Q3 impact at the Head - [133kg carrier robot]
% % Setup C: Dummy Q3 impact at the Legs (Tibia / fibia) - [133kg carrier robot]
% 1 setups with adult dumm HIII (50-percentile adult) [1.77m height / 81.5kg weight]
% % Setup D: Dummy H3 impact at the Legs (Tibia / fibia) - [133kg carrier robot]
analysis_scripts/ :
Data pipelines for modelling, risk assessment and other statistics to operate on the cleaned dataset. Also, it contains visualization functions.
figures/ : Contains visualization for all data, raw and processed by name.
collision_models/ :
Collision models tested and built from the validated data are described here, with examples.
Qolo robot with dummy drive H3 Robot mass: 133 Kg Differential velocity at collision time: [1.0, 1.5, 3.0] [m/s]
Setup A: Collision at chest height. Robot mass: 60 Kg [dotted lines] vs. 133 kg [solid lines] Chest Deflection
Chest Deflection vs. Impact Force ratio
Head Acceleration at the Chest Impact
Head Acceleration at the ground Impact
Setup A: Collision at chest height. Robot mass: 133 kg.
Setup A: Collision at chest height. Robot mass: 60 kg.
Setup B: Collision at head height. Robot mass: 133 kg.
Impact Forces
Head Acceleration at Blunt Direct Impact
Head Acceleration at Ground Impact
Setup C: Collision at Legs height. Robot mass: 133 kg.
Qolo robot with dummy drive H3 Robot mass: 133 Kg Differential velocity at collision time: [1.0, 1.5, 3.0] [m/s]
Impact Forces
Right-leg Tibia Compression Forces
Right-leg Tibia Lateral Forces
Left-leg Tibia Compression Forces
Left-leg Tibia Lateral Forces
Head Acceleration at Ground Impact
References
[1] Paez-Granados D., and Billard, A. “Risks posed by new mobility devices and service robots to pedestrians”. 2021. (Under review)
[2] Paez-Granados, Diego, & Billard, Aude. (2021). Mobile Service Robots Crash Testing with Pedestrians: Safety Assessment with Child and Adult Dummies (1.0) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.5266447.
Contact: [Diego Paez]
Acknowledgments This project was partially funded by the EU Horizon 2020 Project CROWDBOT (Grant No. 779942).