Package detection device

Package detection IoT device

Hardware & Software lab coursework

Do you have package left at the door?

What's the problem?

In our daily life, we often have packages left at the door by delivery men but they usually don’t text or call to tell you that the package has been delivered. Resulting in a higher risk of damage of items and loss of parcels. Having to go out frequently to check for packages outside the door causes a lot of inconvenience.

Customer challenges

To understand our users’ pain points, goals, and motivations, I conducted user interviews with 8 potential customers who have this problem.

Ignore the emails or texts for delivery, forget about their online orders and packages.
Lack of outdoor power supply for monitoring device.
Concerned about camera recording data security and privacy issue.
Existing products like smart home cameras have motion detection function but will affects by passersby/pets/insects movement, and lack of package detection.

How about a device that can recognizes packages periodicaly and alerts users?

Product Overview

The purpose of the device is to detect packages at the door and alert people at home to pick them up in time. The product consists of two parts, display device, sense device and APP that can be updated with monitored information in real time.

Product key features

Customer persona

User story

Acceptance criteria

Powering

As a user without outdoor power supply, I would like to have a monitoring device with a battery.

As a user who traveled for work frequently, I hope this device can independently work for long periods of time without needing to be recharged.

As a user who doesn’t usually check my devices, I’d like the device to show me how much power the device has and remind me to charge or replace the battery.

Remote notification

As a user who doesn’t stay at home very often, I want to receive package notifications remotely when I was not at home.

As a user who doesn’t actively check the monitoring status, I need the device to be able to proactively push notifications when new developments are monitored.

As a user who are busy with family and children, I don’t want to go out to check or setting the device mannually.

As a family user, I hope this device can share notifications to family members remind them of picking package up.

As a co-renting user, I don’t want this device recording or store any data that will violate privacy of others.

Performance & functionality

As a family user who have pets and children at home, I hope this device’s recognition is accurate and will not be interfeced by other objects moving.

As a user live in apartment, the space is commpact so I hope this device is small that do not need much space for settingup and detection.

As a user who shops online frequently, I hope this device can import orders and tracking delivery information automaticaly.

As a user who often loses packages, I would like the device to work 24/7 to monitor that nothing gets missed.

As a co-renting user, I would like the app to be able to differentiate who the packages belongs to by the imported order and notify the corresponding user individually.

As a user who receives tons of notifications and messages every day, I would like the app’s message notifications to be timely, not lagging, and not reminding me of the same information over and over again.

The device should be battery powered.

The device needs to have a highly efficient battery management strategy that allows the device to last more than 3*24h in normal monitoring mode operation and more than 5*24h in energy saving mode operation.
The device should be able to display battery data and operating status in real time in the form of battery life.(for example: Normal Mode – 48h of battery life remaining, Energy Saving Mode – 70h of battery life remaining)

The device should be able to transmit monitoring information to the display device via BLE/Wifi

The device needs to proactively alert the user by 1. Sounding a buzzer on the display device 2. Push message notification to the user’s cell phone through the application.

Device needs to run and reset its state automatically and unattended.

The device application should have the function of group that can create “family” and users can add people into it.

The device should recognization real time capture image and result in 0.5s by ML model on the device instead of storing anything.

The device’s object recognition accuracy should exceed 95%.
Monitoring settings should be configurable to allow users to specify monitoring parameters and sensitivity.

The device should have a threshold for detection range to prevent interference from distant pet/person movement.
The detection threshold should be adjustable depends on users’s space condition.

The application should integrated with API from delivery tracking website.
The app should import order details and tracking information automatically from online purchases.

The device must operate 24/7 to monitor package deliveries and any events. waking up the camera when the trigger condition is reached.
Upon package arrival, the app should notify the corresponding co-renting user individually through their preferred notification method (e.g., push notification, email).
The notification should clearly indicate the recipient’s name and package details for easy identification and retrieval.(e.g. user A, it was detected that you have a package arriving at your doorstep at 12:55am. Click for more package details)
Notifications should not lag behind the event by more than 1 minute under normal operating conditions.

Once a notification has been acknowledged by the user, the app should not repeatedly notify the user of the same information unless there are significant updates or changes.
Notifications should be synchronized across all user devices to ensure consistency and prevent duplicate notifications.

Risks

Schedule risk: There might be sudden changes required in the product feature before the launch, which will be delayed delivering on the initially proposed date.

Prevention – Provide buffer period before the initial launch date so that any unforeseen changes do not affect the product launch.

Budget Risk: Adding any new feature to the product or replacing any selected part model will impact the budget.

Prevention – Spend more time in the testing and validation phase to meticulously anticipate unexpected changes that may occur and the reliability of part selection. Have at least 10 to 20% more additional resources than the proposed budget. Additional resources of 10% to 20% over the proposed budget.

Operational Risk: If there are any changes to the project scope and business objectives, failure to communicate the changes in a timely manner between teams may impact the project.

Prevention – Keeping information synchronized and transparent is key to avoiding losses caused by lagging information and keeping the relevant teams informed of progress on a regular basis.

Technical Risk: Camera-related user data privacy and the risk of cameras being compromised by networks

Prevention – Work with network and hardware security and their associated teams to ensure greater attention is given.

Program risks: legal challenges, communication challenges between end users.

Prevention – Establish a user feedback system and work closely with engineers to iterate on timely improvements for issues that arise during testing and real-world use engineering.

Business Risks: Funding challenges, access to potential customers takes a long time to build brand trust.

Prevention – Conduct as much customer research as possible and design the best scalable business model.

System architecture

Task flow

Device design (pcb)

Data processing

1. Threshold & Denoised
Initialize the detection distance as a threshold value according to the placeable range, and use the moving average method to calculate whether the distance value is lower than the threshold value.

2. Machine learning
Fine tuning YOLOV8 model on custom package dataset to detect package in camera and then segment it.

3.Corner detection
Use cv2 detect the corner of segmentation image

4.Volum calculation
The coordinates of the corner points are assigned using an algorithm based on the corner detection results and the parcel volume size is calculated accordingly.

Segmentation

Train the YOLOV8 model on custom package dataset and segment the package from background

Corner detect

Use CV2 to detect the corner of segmentation image

Volume estimation

Calculate the volume based on the corner's coordinate

GTM strategy

Pricing

We have conducted market research and surveys comparing competitors’ features and services to ensure our pricing is competitive. The pricing range for our home camera products is $60-$120. We also reached out to a diverse group of potential users to gain insight into whether they would be willing to pay for an ongoing subscription to the monitoring feature. and ultimately chose to charge $5.99 per month for this monitoring service on a monthly basis. Users can discontinue sending the device back at any time if there is any dissatisfaction with the process within 12 months.

Marketing

Create marketing content and post video content on social media channels (facebook, instagram) about lost packages, product usage. Integrate the benefits and value proposition of the product in this content, emphasizing the security it provides for parcel safety.

User vitality

In order to improve user experience and trust in our platform, we needed to set up a user feedback system and a 24h support service. In addition we should implement a reward system to encourage existing users to share their experiences and videos on social media.

Product Roadmap

Research, Define and design

focused on thorough market research, defining precise product requirements based on user feedback, and designing a user-centric prototype.

MVP develop

Transformed our product vision into a MVP, integrating core features and undergoing rigorous testing and compliance checks.

Test & iteration

Performance, usability and reliability were optimized through extensive testing and user feedback. Iterative improvements ensure that products meet high standards and increase user satisfaction before they are released to the market.

Sales & marketing

Increased product awareness and adoption through targeted sales campaigns and strategic marketing initiatives. Established strong support and feedback mechanisms to prepare the market for continued growth.