Pedestrian Avoidance System | artificial intelligence in industrial safety | Delaware, USA

Business Impact

  • One of Australia’s Vehicular interaction solution providers approached OptiSol for a PAS (Pedestrian Avoidance System) to protect the staff and operators of the forklifts from harm.
  • Their traditional collision avoidance systems create high numbers of nuisance alerts, which leads to drivers ignoring alerts, which dilutes the value of the system.
  • OptiSol designed a solution that uses Deep Learning to identify humans and calculates their risk in the proximity of forklifts.
  • We used AI/ML to minimize false alarms by evaluating multiple danger factors around the vehicles through various AI/ML algorithms.
  • Our designers also added enterprise features and management capabilities to the solution, notifying managers of incidents that would have gone unreported.
  • Implementing this solution in one of their warehouses reduced the chances of human encounters with forklifts, reduced risks, and improved the safety of the workplace.

OptiSol’s Approach

The PAS devices on the forklifts are connected to the Cloud through IoT Core, which helps manage each device and facilitates data exchange with each device. The whole PAS solution comprises of various AWS services that receive and processes data from the device and provides data and instructions to each device.
The device sends telemetry, state and event data to AWS where the data is scrutinised and decisions are made. Data is also used for reporting and analytics for managers that operate forklifts through the enterprise portal. Managers are prompted with events that are regarded as significant. Images relating to significant events are available for download.

Goal Achieved

Each vehicle has an IoT device and camera(s) installed. This IoT device communicates to AWS IoT Core using MQTT messages. The AWS architecture performs various tasks such as management of each IoT device, upload of images and videos as well as telemetry data. Each device contains accelerometers, temperature, humidity, and many other sensors. This data is fed to the Cloud where AI/ML algorithms derive insights. For example, we can determine if an event has taken place that managers might think of as significant. If so, we send the appropriate manager a message. We can also estimate the maintenance requirements of each vehicle based on the time in use and how hard the vehicle is being used. Managers can access detailed reports that combine all information, including images. For example, managers can download accident or incident information (including images) that can be used in investigations by safety officers or authorities.

Architecture Diagram

Trusted and Proven Engagement Model

  • A nondisclosure agreement (NDA) is signed to not disclose any sensitive information revealed over the course of doing business together.
  • Our NDA-driven process is established to keep clients’ data and IP safe and secure.
  • The solution discovery phase is all about knowing your target audience, writing down requirements, and creating a full scope for the project.
  • This helps clarify the goals, and limitations, and deliver quality products & services.
  • Our engagement model defines the project size, project development plan, duration, concept, POC etc.
  • Based on these scenarios, clients may agree to a particular engagement model (Fixed Bid, T&M, Dedicated Team).
  • The SOW document shall list details on project requirements, project management tools, tech stacks, deliverables, milestones, timelines, team size, hourly/monthly rate cards, billable hours and invoice details.
  • On signing the SOW, an official project kick-off meeting shall be initiated.
  • Our implementation approach, ecosystem, tools, solutions modelling, sprint plan, etc. shall be discussed during this meeting.

Awards & recognition

Awarded Bronze Trophy at CII National competition on Digitization, Robotics & Automation (DRA) – Industry 4.0






Projects for
reputed Clients


5 yrs

in AI & ML

Awarded as Winner among 1000 contestants at TechSHack Hackathon

Related Success Stories

Related Insights

Connect With Us!