Python 3.x Jupyter Notebook to quantitatively identify the airborne collision risk based on the ASTM Committee F38 on Unmanned Aircraft Systems DAA performance standard.
- Airborne Collision Risk Classification
We encourage the use of the GitHub Issues but when email is required, please contact the administrators at encounter-model-ml-admin@mit.edu.
| Acronym | Phrase |
|---|---|
| AGL | above ground level |
| BVLOS | beyond visual line of sight |
| D&R | durability and reliability |
| DAA | detect and avoid |
| JARUS | Joint Authorities for Rulemaking of Unmanned Systems |
| LLSC | Lincoln Laboratory Supercomputing Center |
| SORA | Specific Operations Risk Assessment |
| UAS | unmanned aerial systems |
The continuing integration of unmanned aerial system operations into the National Airspace System requires new or updated regulations, policies, and technologies to enable safe and efficient use of the airspace. Detect and Avoid (DAA) systems utilized by UASs are one such technology that is actively being standardized. To mitigate the risk of midair collisions, these systems consists a functions to sense and identify potential hazardous encounters with other aircraft and then recommend a maneuver such that the UAS avoids the other aircraft.
One standardization effort to define DAA system performance requirement is led by the ASTM Committee F38 on Unmanned Aircraft Systems, Subcommittee F38.01 on Airworthiness. This subcommittee established working group 62668 with the objective to define minimum performance standards for DAA systems applicable to smaller UAS BLVOS operations for the protection of manned aircraft in lower altitude airspace. Smaller UASs are defined in the ASTM standard as having a wingspan or equivalent of 25 feet or less.
This ASTM performance standard prescribes different requirements, such as those for DAA system availability or assurance, based on generalized airborne collision risk classifications for specific operating locations. In response, this software was developed to quantitatively classify and visualize the ASTM-based airborne collision risk classifications for specific locations. The software was not developed to identify the risk a UAS colliding with the ground or third party non-participants. Additionally while the software developers are not a civil aviation authority or air navigation service provider, this software may be used as part of a determination of the initial airborne collision risk classification. This initial determination aligns with many of the concepts described in the JARUS 2019 SORA 2.0.
The ASTM F38 WK62668 DAA performance standard discusses generalized airborne collision risk classifications. These risk classifications were derived from guidance developed by the JARUS 2019 SORA 2.0. It is important to note that these generalized Airborne Collision Risk Classifications were created with mainly cooperative intruders in mind. Care should be taken when using these generalized classifications to describe collision risk with non-cooperative intruders.
While the ASTM standard does not prescribe a set of requirements for each classification nor specify a method to determine air risk, it provides generalized guidance on the context for each risk classification. Given this mostly quantitative guidance, this software use open source datasets to identify and visualize air risk classes given specific latitude, longitude, and altitude positions.
For our definitions, we assume rural areas have a population density of 100 people per square mile or less. While rural areas have not been formally defined by the FAA or for SESAR U-Space, we initially defined rural areas for low risk airspace DAA as having a population density of 100 or less. This quantitative criteria for rural areas was based on the draft D&R MOC from February 2020 or FAA brief to type certificate applicants from July 2019.
As with any generalization, when the area becomes more refined there will be specific areas where the generalized classification levels will be true, and other specific areas where the generalized classification levels will not be true. Care should be taken when using these generalized classifications to describe collision risk with non-cooperative manned aircraft since these classifications were created with primarily cooperative manned aircraft in mind. For specific operations, the UAS operator should work with the local airspace authority to ensure that the appropriate air risk classification is assigned to the operational volume. Furthermore, the software does not estimate nor make assumptions about the likelihood of a UAS encounter a cooperative transponder-equipped or noncooperative manned aircraft. In coordination with the local airspace authority, tt is possible that for a given location, the airborne risk classification could differ for cooperative transponder-equipped and noncooperative manned aircraft.
Specifically, we defined the default risk criteria, within the United States, as the following:
- Uncontrolled airspace
- Below 500 feet AGL
- Over a rural area (100 people per square mile or less, within a US census block group)
- At least 5 nautical miles away from aerodromes
- Uncontrolled airspace (no altitude ceiling)
- Below 500 feet AGL in controlled airspace (Class B, C, D)
- At least 5 nautical miles away from the center point of Class B, C, and D aerodromes
- Below 500 feet AGL over an urban area (greater than 100 people per square mile, within a US census block group)
- Below 500 feet AGL in Class E or G aerodromes
If a location fails to satisfy all the low or medium risk criteria, it is classified as high risk. We assume that these locations are where manned cooperative transponder-equipped aircraft predominately fly and/or where a smaller UAS would frequently encounter cooperative transponder-equipped manned aircraft. Elevation for a given location(latitude, longitude) is obtained by The National Map Elevation Point Query Service (EPQS). If the query does not return a result for the elevation, the AGL altitude can not be calculated. In this event, the location is assigned high air risk by default.
We do not quantitatively define or identify locations where manned aircraft manned aircraft predominately do not fly and/or were it is extremely improbable for a smaller UAS to encounter any manned aircraft.
This section specifies the run order and requirements for the initial setup of this repository. Any changes to third party software or dependencies shall be in compliance with the respective licenses.
If you have not already, complete the initial setup associated with the em-core repository. This setup will download some of the third party data files, such as the FAA airport dataset, that is used by this repository.
Anaconda can be downloaded on Mac, Windows, or Linux. Select the python 3.x version. Once downloaded, to create an identical Anaconda environment to the one used in developing the code, run the code below to install all dependencies.
conda env create -f environment.ymlOnce downloaded, to activate the Anaconda environment you will either need to run source activate airspace or conda activate airspace from the command line depending on which version of Anaconda is installed. Specifically, first activate your anaconda environment by running either:
source activate airspaceor
conda activate airspaceNote that older versions of Anaconda use source activate <env name> while newer versions use conda activate <env name>.
This is a set of boilerplate scripts describing the normalized script pattern that GitHub uses in its projects. The GitHub Scripts To Rule Them All was used as a template. Refer to the script directory README for more details.
You will need to run the following script to download and process the Census block group shape files.
This repository contains processed U.S. Census population density data for the year 2019. If different census data is desired, you need to reprocess the U.S. Census data. The first step is to sign up for a U.S. Census API key. After signing up, ensure that you have copied the contents to the API.txt file on your local clone of this repository. You can then process the new data by running the following lines from within the scripts directory:
python python/block_group_process.pyIt is essential to make sure you are in the scripts directory before running the line above! If you are in the root directory of the repository, this can be achieved by running cd scripts/.
The run order assumes that the Anaconda airspace environment is activated (conda activate airspace), as prescribed in the initial setup.
You should now be able to run the contents of this repository, including Main.ipynb from a Jupyter notebook. First launch Juypter:
jupyter notebookThen click on the Main.ipynb notebook that you want to run. If you are unfamiliar with the Jupyter notebook interface, please refer to this nice introduction and tutorial. In our walkthrough of the notebook, we specify a state and then create a longitude/latitude grid of points to run the airspace risk classification. This code does not depend on the state and longitude/latitude/altitude points can be read in from a file (or created from a different geometric boundary) to run the classification on. A conveinience script (runState.py) is provided for running the airspace risk classification for an entire state.
If for some reason Jupyter is unable to launch or you are receiving errors with the environment, one solution is to try reinstalling the Jupyter module. This can be done by first activating the environment conda activate airspace, and then running the install command conda install jupyter.
If you want to simply run the code to see the output, first click on one of the code cells. Then press shift-enter to run the code cell. You will run shift-enter for each code cell that you want to run. If you want to just run all cells, click on kernel at the top, then restart & run all. Refer to Jupyter notebook documentation for more guidance on how to use manipulate notebooks.
If you would like to run the airspace risk classification for an entire state without the use of the Main.ipynb. The script runState.py is provided for this purpose. Activate the airspace Anaconda environment and run:
python runState.py STATE SPACING_deg alt_ft_aglWhere STATE, SPACING_deg, and alt_ft_agl are arguments that you can pass to the script. Here is an example for processing the airspace risk class for North Carolina with 0.1 degree spacing and at 500ft agl.
python runState.py NC 0.1 500The output dataframes will be saved to the output/ directory. To organize the output, a sub-directory with the corresponding SPACING_deg value will be created to store the dataframes from runState.py. From our example above, the dataframes will be found in output/spacing/0.1/.
Here we define the steps within the Main.ipynb notebook provide an example configuration for processing the airspace risk class for an entire state.
The following variables will need to be defined in Main.ipynb to run the airspace risk classification (see Main.ipynb for the location of where the variables are set):
| Variable | Type | Description |
|---|---|---|
path_to_emcore |
string | (required) Contains the path to the em-core directory. The code will automatically try to grab the path from the system environment variables. If it is not there, the user will be prompted to enter this information. |
SPACING_deg |
float | (optional) This variable defines the spacing in degrees between longitude and latitude points when generating a uniform grid over a state (or polygon). The default is 0.01. Note SPACING_deg ~0.02 degrees is equal to 1-1.2 nautical miles. |
STATE |
string | (optional) This variable defines the state to process when performing the airspace risk classification over an entire state. The default is NC. |
The next step after defining variables in Main.ipynb, is to load the required open source data. The required data is as follows:
| Data | Description | Download Location |
|---|---|---|
| FAA Airports | Airport geometry shapefiles. | em-core initial setup. |
| FAA NASR | Airspace class information, along with their corresponding geometries. | em-core initial setup. |
| NE-Admin | Geometry shapefiles for all states in the US. | em-core initial setup. |
| Census Block Group | Population density values and geometries for all US Census block groups. | scripts/setup.sh. |
Next, we are ready to begin classifying the airspace risk class for the longitude/latitude points. In Main.ipynb, code blocks [4] and [5] detail a preprocessing step for generating a list of longitude/latitude points within a particular state. Code blocks [4] and [5] can be skipped if you already have a list of longitude/latitude points, such as position reports from an aircraft track. The format of the input longitude/latitude array is as follows:
| Longitude 1 | Latitude 1 |
|---|---|
| Longitude 2 | Latitude 2 |
| ... | ... |
| Longitude N | Latitude N |
The main function call that classifies the air risk class for the input longitude and latitude points is listed below:
| Function | Description | Inputs |
|---|---|---|
| RiskClassification | Performs the airspace risk classification for the input longitude/latitude points. | [lonlats, data, alt_ft_agl (optional), points(optional)) |
lonlatsis a required input into theRiskClassificationfunction. It is the array of longitude/latitude points that will be classified as a either Low, Medium, or High risk airspace.datais a required input into theRiskClassificationfunction. It is a class containing all of the required data that is listed in Loading Dataalt_ft_aglis an optional input into theRiskClassificationfunction. If there are associated altitude values with the longitude/latitude array, the altitude can be provided as long as it contains the same number of data points as the longitude/latitude array. If this field is not specified, a default value of 500ft AGL is assigned to all longitude/latitude points.pointsis an optional input into theRiskClassificationfunction. It represents a geometric point for each longitude/latitude point which is used throughout theRiskClassificationfunction. If it is not specified, it will be automatically created.
The function can be called as follows:
df,lr_df = RiskClassification(lonlats,data,points=points)RiskClassification returns two outputs: df and lr_df. df is a data frame that contains the airspace risk classifications in three categories: Low, Medium, and High risk. lr_df only provides two categories of classification: Low Risk and Not Low Risk.
The fields of df are displayed below:
| Field | Type | Description |
|---|---|---|
| geometry | geometry (shapely) | longitude/latitude point |
| Risk_Class | float | Integer representation of airspace risk class. 0: Low Risk, 1: Medium Risk, 2: High Risk |
| Status | string | Description of why a point was classified to a particular risk class, (e.g. what criteria did the point fail?) |
| Type | string | String representation of airspace risk class. LR: Low Risk, MR: Medium Risk, HR: High Risk. |
Once the data frames are obtained we can plot the airspace risk classification. Details on saving the output can be found in Main.ipynb. Below is the output of Main.ipynb for three different states with SPACING_deg=0.0035.
STATE='NC'
STATE='VA'
STATE='KS'
Please use the appropriate documents listed in em-overview/README when citing the technical concepts.
Please use this DOI number reference when citing the software:
The Lincoln Laboratory Supercomputing Center was used by MIT LL to process the data. When referencing the LLSC, please use this article:
A. Reuther et al., "Interactive Supercomputing on 40,000 Cores for Machine Learning and Data Analysis," 2018 IEEE High Performance extreme Computing Conference (HPEC), Waltham, MA, 2018, pp. 1-6, doi: 10.1109/HPEC.2018.8547629.
@inproceedings{llsc2018,
author={A. {Reuther} and J. {Kepner} and C. {Byun} and S. {Samsi} and W. {Arcand} and D. {Bestor} and B. {Bergeron} and V. {Gadepally} and M. {Houle} and M. {Hubbell} and M. {Jones} and A. {Klein} and L. {Milechin} and J. {Mullen} and A. {Prout} and A. {Rosa} and C. {Yee} and P. {Michaleas}},
booktitle={2018 IEEE High Performance extreme Computing Conference (HPEC)},
title={Interactive Supercomputing on 40,000 Cores for Machine Learning and Data Analysis},
volume={},
number={},
pages={1-6},
month = july
year={2018},
doi = {10.1109/HPEC.2018.8547629},
url = {https://doi.org/10.1109/HPEC.2018.8547629},
}The National Map Elevation Point Query Service courtesy of the U.S Geological Survey is used when calculating the AGL altitude for a given latitude and longitude coordinate.
© 2020 Massachusetts Institute of Technology.
This material is based upon work supported by the Federal Aviation Administration under Air Force Contract No. FA8702-15-D-0001.
Delivered to the U.S. Government with Unlimited Rights, as defined in DFARS Part 252.227-7013 or 7014 (Feb 2014). Notwithstanding any copyright notice, U.S. Government rights in this work are defined by DFARS 252.227-7013 or DFARS 252.227-7014 as detailed above. Use of this work other than as specifically authorized by the U.S. Government may violate any copyrights that exist in this work.
Any opinions, findings, conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the Federal Aviation Administration.
This document is derived from work done for the FAA (and possibly others), it is not the direct product of work done for the FAA. The information provided herein may include content supplied by third parties. Although the data and information contained herein has been produced or processed from sources believed to be reliable, the Federal Aviation Administration makes no warranty, expressed or implied, regarding the accuracy, adequacy, completeness, legality, reliability or usefulness of any information, conclusions or recommendations provided herein. Distribution of the information contained herein does not constitute an endorsement or warranty of the data or information provided herein by the Federal Aviation Administration or the U.S. Department of Transportation. Neither the Federal Aviation Administration nor the U.S. Department of Transportation shall be held liable for any improper or incorrect use of the information contained herein and assumes no responsibility for anyone’s use of the information. The Federal Aviation Administration and U.S. Department of Transportation shall not be liable for any claim for any loss, harm, or other damages arising from access to or use of data or information, including without limitation any direct, indirect, incidental, exemplary, special or consequential damages, even if advised of the possibility of such damages. The Federal Aviation Administration shall not be liable to anyone for any decision made or action taken, or not taken, in reliance on the information contained herein.
This product uses the Census Bureau Data API but is not endorsed or certified by the Census Bureau.