Prototype Model
Final Prototype (With camera and components inside)

🧸🤖 EchoCare Bear

🔎Problem Statement

How might we leverage emerging technologies to reduce social isolation within the elderly community in Singapore.

🤯 Inspiration

2/5 elderly individuals in Singapore experience social isolation. While current solutions like apps and online platforms encourage social interaction, they often fall short as elderly individuals may forget to engage with the system. In 2023 alone, there were at least 37 cases of ‘lonely death’ in Singapore, with roughly 40,000 in Japan as of August 2024.

Additionally, the proportion of older adults (aged 65 and above) living alone has increased steadily over the past two decades. This trend is driven by contributing smaller family sizes, children moving out or residing overseas for work, and evolving social norms around multi-generational living. These factors, coupled with issues like loneliness, fall-related accidents, and lack of mental health support, make elderly care a pressing societal concern, which inspired us to create a more proactive, empathetic, and autonomous AI companion ⸺ EchoCare.

🚀What It Does

EchoCare takes inspiration from Lion Befriender's I’mOK app, which ensures that elderly individuals living alone are safe.

EchoCare tackles the pain points faced by the elderly population, particularly those living alone. It offers the following functionalities:

1.Well-Being Check-In

Elderly individuals often experiences loneliness and lack of regular check-ins, leading to increased mental health issues among elderly. Hence, EchoCare would:

Checks the well-being of the elderly every morning and evening.
Uses casual small talk to build rapport.
Responds empathetically when detecting distress or discomfort.

2.Regular Reminders

Elderly often forget to take their medication and attend events due to their declining memory that comes with age. Hence, EchoCare would:

Provides daily reminders for medication, hydration, and daily routines.

3.Emergency Detection

Many a time, elderly do not have immediate access to emergency services, and delayed responses could lead to severe consequences in their health. Hence, EchoCare would:

Alerts emergency services if the elderly fall or fail to respond to a check-in.

4.Personalization

Generic systems lack empathy and personalisation, making elderly more resistant in using technology. Hence, EchoCare would:

Maintains a friendly and soothing tone.
Takes on the persona of a caring friend or caregiver.

5.Therapy-Like Interaction
Elderly may face emotional distress or anxiousness without immediate access to mental health services, and their reluctance to get help from others exacerbates the problem. Hence, EchoCare would:

Provide autonomous emotional support
Offers calming statements if the elderly seem upset or anxious.
Escalates to a caregiver if the situation appears serious.

🛠️ How We Built It

For the hardware, we utilized a Raspberry Pi 4 Model B, motor drivers, a motor controller, chassis wheels, wooden boards, and a Teddy Bear from ESPRESSIF. For the software, we leveraged Qwen 2 Audio and Python as our primary programming language.

😰 Challenges We Encountered

Limited access to 3D Printing: We had to adapt our design and rely on a laser cutter and wood due to restrictions on 3D printing.
ESP32-S3-Box-3 Integration: The unfamiliarity with this device and its built-in library functions made it challenging to incorporate into our design, especially with all of our team members using different OS systems
Low-Quality Motor Drivers: The motor drivers were cheap ones and difficult to work with
Restrictions due to Raspberry Pi: Trying to run the LLM on the raspberry pi proved to be difficult due to its computational powers.

🎯 Accomplishments We’re Proud Of

Successfully integrating a large language model (LLM) to enable meaningful communication with the elderly.
Adapting our design to overcome the limitations imposed by the hackathon's rules and restrictions.

✏️ What We Learned

Expect the Unexpected: We realized that in fast-paced projects, everything that can go wrong often does, and adaptability is key.
The Value of Participation: Hackathons are not just about building solutions but also about enjoying the creative process and the collaborative atmosphere.

🤔 What’s Next for EchoBear?

Adaptive Learning: Enhance the robot’s capabilities by learning specific elderly habits, such as meal times or wake-up schedules, to provide a more personalised and catered experience.
Implement Computer Vision: Implement path finding using OpenCV such that the robot is able to autonomously follow the elderly and detect any emergency situations (such as the elderly falling down)
Community Campaign: Launch a community initiative to raise awareness about the dangers of extreme social isolation among the elderly. Through user testing and feedback and using Data Analytics, refine EchoCare Bear to better serve its purpose of improving elderly lives.