THE ORIGINAL CONCEPTION, Ball-E V1.0
DESIGN SPECIFICATION
Our original design aimed to create a caregiver robot specifically designed for toddlers. The primary objective was to develop a mobile CCTV camera that could provide additional support to primary caregivers and keep parents updated with real-time images and videos of their child, even while they were at work. Ball-E V1.0 was initially envisioned to incorporate machine learning algorithms, enabling it to identify potential dangers to toddlers and promptly notify caregivers and parents via SMS alerts.
The movement mechanism of Ball-E V1.0 involves the rotation of its outer case, which consists of two transparent plastic hemispheres. A main wheel powered by a DC motor at the bottom propels it forward, while four lateral wheels, driven by stepper motors, provide the ability to change directions smoothly. To ensure stability, the support structure is strategically designed to incorporate two heavy Lithium-polymer batteries, effectively lowering the center of gravity and allowing Ball-E V1.0 to maintain an upright position. Additionally, an embedded camera (in yellow) is incorporated into Ball-E V1.0, which, when coupled with machine learning algorithms, enables it to monitor the activities of a toddler with great precision and accuracy.
During the initial stages of planning it was thought that the cute appearance, lack of sharp edges and small external parts of a spherical robot would make it an excellent toddler monitoring system.
CHALLENGES AND FEEDBACKS
01 CAREGIVING
Providing effective caregiving to children is a multifaceted responsibility that demands quick and accurate responses. Meeting the dual requirement of accurate hazard identification and prompt reporting to parents and caregivers to prevent accidents is crucial for ensuring child safety. However, Ball-E faced challenges in fulfilling these demands effectively.
02 ALGORITHM
Our initial tests focused on assessing the algorithm’s capability to track a moving human subject. However, the motion tracking functionality did not perform as intended, and the algorithm frequently lost track of the moving target after a few seconds, particularly when the person turned away from the camera.
The limitations of the image recognition algorithm posed significant obstacles to Ball-E ’s ability to accurately monitor and track potential hazards. Without consistent and reliable tracking, the robot’s effectiveness in promptly notifying parents and caregivers of potential dangers was compromised.
MOVING FORWARD
With the difficulties experienced, feedbacks from our mentor and further team discussion, our team agreed on incorporating the following changes to Ball-E:
- Devise a less sophisticated application of Ball-E other than caregiving, which was a complex task.
- Redesign the motors within Ball-E to reduce total weight to increase its potential applications, and reduce total number of components required for its construction.