• Verify sidewalk location• Update 10-ft. sidewalk width• Update 10-ft. cross slope• Update 40-ft. grade
• Verify curb ramp locations• Update curb ramp approach slope
Improving Pedestrian Infrastructure Inventory in Massachusetts using Mobile LiDARChengbo Ai, Principal Investigator; Qing Hou, Graduate Research Assistant
Department of Civil and Environmental Engineering, University of Massachusetts Amherst
Problem Statement
Pedestrian infrastructure is one of the most vitalinfrastructures for pedestrians and wheelchairusers to facilitate safe and uninterrupted trips intheir everyday lives. To meet the obligations ofthe ADA Transition Plan, and to make informedinvestment decisions in transportation assetmanagement, MassDOT is responsible for theirtimely maintenance of inadequate sidewalks inits jurisdiction. Through its goals and policies,MassDOT is seeking to provide all modes oftransportation with equitable accommodations.
Methodology
Additional Information
Results and Findings
However, the current sidewalk inventory dataresides within MassDOT is not of a quality orlevel of detail to make informed decisions. Theprevious effort has been made to inventory thecritical pedestrian infrastructure. However, mostof the effort still included a large amount ofmanual process, and yet did not take advantageof the widely available LiDAR data.
This research project seeks to demonstrate andevaluate the feasibility of mobile LiDAR system asa viable technology to facilitate cost-effectiveinventory update and condition assessment forpedestrian facilities within the Commonwealth.With the identified locations and the embeddedmeasurements, the derived database from themobile LiDAR can be integrated with the existingroad inventory seamlessly, which will provideMassDOT with accurate information from whichto prioritize sidewalk maintenance needs.
Automated LiDAR point cloud segmentation
Objective
Automated sidewalk and curb ramp extraction
The proposed method demonstrated that mobile LiDAR iscost-effective for network-level pedestrian infrastructureanalysis through a case study, covering the entire StateRoute 9 corridor (271.76 miles)
• Processed more than 8 billion LiDAR points
• Extracted 85 miles of sidewalks and conducted thecorresponding measurements at 7 min/mile
• Extracted/updated 1,297 curb ramps and conductedthe corresponding measurements at 2.2 min/mile
This project was conducted as part of the MassDOTResearch Program with funding from FHWA SPR funds. Formore information, please contact Dr. Lily Oliver, Managerof Research, ([email protected]), and Dr. ChengboAi, Principal Investigator, ([email protected]).
The research team also developed a complete point clouddata processing pipeline (e.g., tools, algorithms, GUIs andprocedures) that can facilitate the implementations of astatewide pedestrian infrastructure inventory and manyother transportation assets inventories.
The point cloud segmentation method employs the deep learning-basedapproach, i.e., PointNet++, to decompose the large-scale, raw point clouddata into feature-rich, meaningful groups so that the subsequent sidewalkextraction algorithms can be applied efficiently.
The sidewalk extraction method employs an efficient coplanar identificationapproach using Principal Component Analysis and Octree. The algorithm issequentially applied to 3-ft. LiDAR stripe along the driving direction.
The curb ramp extraction method employs an image-based, DeformablePart Model to identify the individual parts of a curb ramp, and then projectthe results to the LiDAR based on tight image-LiDAR registration.
The final report of this research project is available athttps://www.mass.gov/lists/current-and-completed-research-projects
