​Objectives/Project/Challenge:

• The City of Scottsdale Department of Transportation (DOT) wants to collect data on pedestrian counts and safety by assessing the movements on roads, sidewalks, and curbside utilization. The city can use the data and analytics to understand pedestrian safety, plan capital projects, and design planning. 
  
• Use Cases:
  
  • Pedestrian counts
    
  • Pedestrian movements
    
  • Near-miss collisions involving pedestrians
    
• Objectives
  
  • Understand the counts and risks of pedestrians, cyclists, and other mobility road users' vehicle engagement.  
    
  • Data can be used to support obtaining capital projects and development.
    

The Solution:

• The solution is out-of-the-box with a camera or LiDAR sensors; an environmental enclosure is preassembled with an IoT edge device and an LTE router with a sim card. No connection to the city’s network is required as our IoT edge devices are connected to an LTE 4G router. Sensagrate is a Verizon partner and connects to their LTE 4G network. If the 4G connection is impossible or the selected location has weak signal solutions, Sensagrate will work to find alternative solutions. 
  
• We leverage existing infrastructure to deploy SensaVision Edge on the traffic control signal, street light, or utility poles with an accessible power source for a successful deployment. SensaVision is powered through luminaires using a streetlight power adapter. To power the SensaVision Edge system through an enclosure that has 120 V AC (with two (2) outlets), and SensaVision Edge uses ~ 84V DC per deployment.
  
• For this project, Sensagrate collected data to assess pedestrian, cyclist, and vehicle movement for analysis to improve pedestrian and cyclist safety through predictive analytics and safety improvement recommendations. SensaVision Edge has an IoT monitoring and reporting system managed and maintained by Sensagrate. The IoT system is secured, and the edge device is encrypted for all data transactions. Sensagrate tracks, addresses, and supports all issues based on our Service Level Agreement to provide timely maintenance to resolve any pro
  

Results:

• During the data collection period, we counted 29,276 pedestrians in the southbound facing camera and 38,064 pedestrians in the westbound facing camera. We identified that the following days had the highest daily volume in this order. We observed that the daily traffic volume counts are the same for vehicles and cyclists. 
  
• We detected 2,664 PET near-miss collisions in the westbound (WB) camera, with 252 involving pedestrians. Of the 252 PET near-misses with pedestrians, thirty-three (33) were severe near-misses. We measure severe PET near-misses as a vehicle going above 30 miles per hour (mph) with a severity rating of 0 to 1.5. Of the 33 severe PET near-misses, nine (9) (or 0.34% of total PET near-misses) involved vehicles above 30 mph.
  
• We detected 1,830 PET near-miss collisions in the southbound (SB) camera, with 321 involving pedestrians. Of the 321 PET near-misses with pedestrians, thirty-eight (38) were severe near-misses. Of the 38 severe PET near-misses, five (5) (or 0.27% of total PET near-misses) involved vehicles with speeds above 30 mph.
  
• We detected 3,719 TTC near-miss collisions in the westbound camera, with 514 involving pedestrians. Of the 252 TTC near-misses with pedestrians, fifty-eight (58) were severe TTC near-misses. We measure severe TTC near-misses as a vehicle going above 30 miles per hour (mph) with a severity rating of 0.0 to 0.6. Of the 58 severe TTC near-misses, twelve (12) (or 0.32% of total TTC near-misses) were at vehicles above 30 mph.
  
• We detected 2,225 TTC near-miss collisions in the southbound camera, with 361 involving pedestrians. Of the 361 TTC near-misses pedestrians, forty-two (42) were severe TTC near-misses. Of the 42 severe TTC near-misses, seven (7) (or 0.27% of total TCC near-misses) were at vehicles above 30 mph. 
  
• We overlaid the tracking objects onto the field of view of cameras to assess the common and unique behaviors of pedestrians on a heat map, thereby identifying the common path areas taken by pedestrians. The color representation with more points will be red, orange, and yellow. The darker the color, the more it represents common trajectory paths. We identified that not all paths are in the crosswalk; during late times in the early morning, with no vehicles, people would cross diagonally at the intersection.
  
• We overlaid the near-miss events of PET and TCC, showing where they occur with the pedestrian trajectory data. With the trajectory paths and near-misses, you can identify locations of high-density near-misses.
  

After pandemic of 2020 to 2021 the campus reopened and allowed e-scooters back on campus. This meant both privately owned and mobility sharing solutions. The Director of Transportation at the university wanted to identify the human behavior on campus at a major intersection. This included safety, speed, location, and movements. 

To meet these objectives Sensagrate installed its SensaVision Edge solution with camera, LiDAR and Radar sensors connected to a roadside unit (RSU) at an intersection with LTE connective to send data in real-time to the cloud and provided data and analytics through the SensaVision web page and report. 

The Results

• For over 120 days we used camera and LiDAR sensors with a 360 degree field of view and range up to 100 meters plus to detect, classify and track e-scooters, pedestrians, cyclists, and vehicles.
• We obtain counts and directions of e-scooters 24/7 (note since the site was a part of research lab the university there were gaps of days through our data collection).
  • Counts included hourly, daily, and weekly counts.
  • The university was able to understand frequency and utilization of e-scooters at a major intersection.
• We captured near-miss collisions analysis for Post-Encroachment Time (PET) and Time to Collision (TCC). The engagements included e-scooters and vehicles, e-scooters and e-scooters, e-scooters and bicyclists, and e-scooters and pedestrians. 
  • The university's transportation department was able to meet initiatives in apply solutions to meet Vision Zero. Understand near-miss collisions frequency involving all vulnerable road users, the risks, mobility objects involved, and timing allows others at the university to learn, understand and play for a smarter and safety mobility on and around campus. 
• For tracking the movements and behaviors we found e-scooters rode in the wrong direction of traffic, violated traffic signal and rightway compliances, rode between people and cars, and did not use bike paths and in some cases bike paths were not available for them.
  • Following our feedback the university implemented a bike path on the SE corner of the intersection. 

Key Takeaways

• Real-time automated counts of e-scooters, pedestrians, cyclists, and videos with a 95% accuracy counts. 
• Detection of e-scooters.
• The university gained an understand of counts and the paths e-scooters are using at a major intersection on campus. 
• The path behavior lead to a safety initiative to add a bike lane for shared multi-modal mobility. 
• The advantage of using SensaVision to perform 24/7 data collection and analysis across multiple locations on campus.
  • Expand to 5 additional sites on campus in 2024. 
  • Applying for USDOT grants to support initiatives for Vision Zero. ​

Learn more, request a demo.