Docilis Engineering Log // ARCHIVE

Lab Notes.

Detailed technical publications on embedded AI, computer vision mathematics, and autonomous robotic embodiment.

DISPATCH #01 // Product Launch / Applied AI•August 2026•~3 min read

#Real-World Usability#Edge Compute#Safety Preemption#DVIE v1.5

Announcing DVIE v1.5 - Real-World Intelligence Beyond the Cloud

Taking the lessons learned from our initial prototype and rebuilding the core into a hardened, production-ready edge daemon.

Today, we are announcing Docilis Visual Intelligence Engine (DVIE) v1.5, the software architecture powering Project AI4Sight. Here is how we engineered smart interruption audio queues, 5-zone spatial grids, and zero-cloud privacy.

Author: Luke Tsalwa (Founder & Lead Researcher)

Read Full Dispatch

DISPATCH #02 // Computer Vision / Edge Engineering•July 2026•~4 min read

#ByteTrack#EMA Smoothing#Ping-Pong Bug#Spatial Memory

Lab Notes - Teaching AI Object Permanence & Motion

How we eliminated the Ping-Pong bug using ByteTrack spatial association and Exponential Moving Average temporal smoothing.

Picture setting your mug down next to your laptop. Your brain recognizes both are still. Here is how we solved the Ping-Pong oscillation bug and camera jitter on edge vision hardware.

Author: Luke Tsalwa (Founder & Lead Researcher)

Read Full Dispatch

DISPATCH #03 // Embedded Systems / Edge Hardware•June 2026•~4 min read

#Bare-Metal ARM#Hardware Fault Tolerance#Systemd Resilience#NCNN

Lab Notes - Squeezing Heavy AI onto a $100 Chip

Bare-metal ARM optimization with NCNN, hardware fault tolerance, and self-healing camera loops.

Building high-tier AI software inside a cloud lab with unlimited compute is one thing. Making it run on a $100 credit-card-sized board on smart glasses is a completely different engineering challenge.

Author: Luke Tsalwa (Founder & Lead Researcher)

Read Full Dispatch