Assignment Taken

Detailed overview of IT infrastructure for the company

Case Understanding

Our company has a vast reach in India and its looking to venture into hyperlocal delivery segment. The hyperlocal services market size was valued at $1,324.2 billion in 2019, and is estimated to reach $3,634.3 billion by 2027, registering a CAGR of 17.9% from 2021 to 2027. For on-demand delivery business models, smartphones have become a game-changer. The rise of on-demand apps has made our lives easier than ever before, from buying groceries to ordering food, purchasing medicine to renting furniture. Consumers now prefer to have all of their basic needs met at their doorstep in order to save time and relax while waiting for their order to arrive. There is a rising demand for fulfillment of the delivery of this sector and the company can leverage its position and presence to capitalize on this opportunity. The company has identified the use of short distance drones due to low initial investment. The company aims to break even within its first year of operation and hence we need to have a solid customer base with good profit margins to achieve this target.

BCS Solution Summary

Developing AI technology for Drone delivery system

Solution

IT & ANALYTICS DEPARTMENT

 

AI for Autonomous Delivery of materials through Drones:

SOFTWARE PLATFORM

 

A proprietary software platform needs to be developed in fulfilling the needs of local staff in activating and monitoring the delivery. 

The platform requires:

1. Backend
2. Application(APP)
3. AI

This whole delivery process helps in improving the efficiency and waiting time. To be guided to activate the autonomous flight, simply select the appropriate route from the app menu.

The Backend allows further monitoring of the different missions performed by different systems running in parallel in a real working scenario.

• WEB PORTAL: A tool for using the administration functions of the platform, monitoring and management of process (backend user)
• MOBILE APP: A tool used for functionalities of request, authorization and reception of transport in mobility
• API LAYER: Architectural component responsible for displaying services necessary for the implementation of user functions 
• APPLICATION LAYER: Application logic for the management of function systems in compliance with the designed logics
• PUSH NOTIFICATIONS: form for sending alerts and notes on mobile devices
• INTEGRATION LAYER: Module for the development of integration logics with external systems and the connectivity module
• CONNECTIVITY: Sensor management logic ( measurement acquisition and transmission of the same) and the control of the drone
• AI : The software running on the special package used as drone carriage for flight activation and supervision
• DRONE: The software running on the commercial flight controller of the drone

 

 

 

HOW AI PLAYS AN IMPORTANT ROLE IN DRONE DELIVERY SYSTEM:

 

OBSTACLE DETECTION AND COLLISION:

Application of AI: Developing Onboard computer vision software that allows drones to detect obstacles and avoid them even at high speeds. Several sensors are placed on all sides of the drone to spot things like oncoming aircraft and other barriers and to constantly monitor the drone’s flight.

GPS FREE NAVIGATION:

Application of AI: They are used to navigate complex spaces. These autonomous drones build a map of their surroundings while tracking their movement through that environment. As a result, drones are completely self-reliant for open-ended exploration and do not require human interaction during the flight. 

CONTINGENCY MANAGEMENT AND EMERGENCY LANDING:

Application of AI: In the event of an emergency or a change in the weather, AI-enabled live data processing systems make predictions based on variables like the drone's current position, energy state, altitude, and wind speed, and then apply an algorithm to steer the drone to a safe landing site or hover in the same area until conditions change. In the case of a forced landing due to system or flight failure or emergency situations, image sensors and AI-enabled object identification software can assist the drone in finding a secure landing spot.

DELIVERY DROP:

Application of AI: The drone's computer vision, in combination with LiDAR and other sensors, can determine how far away it is from the ground and detect if anything is obstructing its ability to safely lower the cargo to the ground.

SAFE LANDING:

Application of AI: Drones with AI-enabled sensors can track their position and speed and adjust their landing trajectory and rotor speed to make the smoothest landing possible. Drones can land without the worry of flipping or damage thanks to image sensors, AI-enabled object recognition, and drone flight computers.

 

 

CHALLENGES FACED DURING AI AND COMPUTER VISION DATA QUALITY IN DRONES:

 

Large amount of data is produced from cameras and sensors which are present on various drones. So, labeling data to ensure a safe model is also a highly technical.

Drone needs do follow process to process the data:

1. Pre-processing drone data:

Drone data can be pre-processed to produce accurate duplicates of real-world objects like machines, rooftops, and automobiles. A mix of 2-D photos, lidar sensors, and photogrammetry can be used to create 3-D renderings of the environment when employing drones for inspection reasons, a process known as digital twin production.

Digital twins serve as a link between the real and virtual worlds. For example, in the drone delivery industry, digital twins can teach one pilot to operate one drone and another pilot to run several drones before transitioning to a drone dispatch model without a pilot. In the digital realm, performing these types of experimental training programmes carries a lower risk and costs less than in the physical world.

1. Live processing drone data:

Live-processing improves object identification, allowing the drone to make game-time decisions like avoiding other planes, flying around weather, avoiding obstacles, and landing safely. If the drone employs simple sensors, they may appear to be the same, but computer vision distinguishes the lakes and adds limitless value to operations.

1. Post-processing drone data:

Drone data post-processing can greatly increase drone delivery safety and efficiency. Sensor, picture, and video data, for example, might indicate the severity of weather difficulties a drone may encounter during flight. Assume that evidence indicates that high winds during a specific season have an impact on battery life or performance. Operators can then change the route or make other changes to the flight plan. Post-processing can also assess whether pre-determined flight paths are the most cost-effective based on demand. If this is the case, flight trajectories can be altered.

SOLUTION:

People:

Labelling of AI data is most important decision. Data from computer vision and sensor learning model is only the good data that people are trained on. Need people to prepare and quality check of images, video and sensor annotations.

Process: 

Designing a computer vision model for drones is an iterative process. Data annotation evolves as you train, validate, and test your models. Along the way, you’ll learn from their outcomes, so you’ll need to prepare new datasets to improve your algorithm’s results.

Your data annotation team should be agile. That is, they must be able to incorporate changes in your annotation rules and data features. They also should be able to adjust their work as data volume, task complexity, and task duration change over time.

Your team of data annotators can provide valuable insights about data features - that is, the properties, characteristics, or classifications you want to analyze for patterns that will help train the machine to predict the outcome you’re targeting.

Technology: 

Data annotation tools are software solutions that can be cloud-based, on-premises, or containerized. You can use commercial tools to annotate pipelines of production-grade training data for machine learning or take a do-it-yourself approach and build your own tool.

 

IMAGE ANNOTATION TOOLS:

 

1. Bounding Box
2. Tracking
3. Polygon
4. Polyline
5. 3-D Cuboid
6. 2-D and 3-D semantic Segmentation

Conclusion

Setting up IT infrastructure for drone delivery system is a crucial part

Attached File Details

View Attachment file (Drone delivery system.png)