Design
Turn the game problem into a testable mechanism, program, strategy, or model.
Becoming Frontier // 2026-2027
What we build.
How we learn.
This is our public record of the OtterBots BIOBUZZ journey: human ambition directed through the Design, Measure, Analyze, Improve, and Control loop—with Microsoft Scout and our LLMWiki helping evidence compound.
Engineering process
Design becomes controlled learning
We do not treat the first working idea as the final answer. Every major choice should preserve the alternatives, evidence, and lesson that produced it.
Measure
Capture repeatability, timing, accuracy, failure modes, and test conditions.
Analyze
Find the bottleneck, contradiction, or assumption that the evidence no longer supports.
Improve
Change one meaningful variable and retain only progress that survives another test.
Control
Make the result repeatable through feedback, documentation, training, and safe limits.
Robot architecture
Start simple, leave room to grow
Our current software foundation supports normalized mecanum driving, operator control, telemetry, and camera-guided AprilTag work. We use that foundation to evaluate autonomous movement and repeatable HIVE alignment without hiding the logic from students.
- Readable drivetrain mixing and normalization
- Safe stop behavior and live telemetry
- AprilTag cluster geometry and distance calculations
- TeleOp alignment and autonomous feedback loops
- Timeouts and explicit state transitions
Control systems
Feedback should change behavior
Our vision programs turn camera observations into robot actions. AprilTag detections are converted into distances, target centers, steering corrections, and stopping decisions. Each step is exposed in the Blocks diagrams and explained in plain language.
ObserveRead visible AprilTags and camera-space measurements.
CalculateEstimate 3D distance, rotated offsets, cluster midpoint, and bearing.
ActTurn and approach using bounded motor commands.
ProtectStop on target loss, timeout, completion, or OpMode end.
Our evolving toolchain
Tools change. The learning loop remains.
Each tool entered the workflow because it removed a measured bottleneck: finding evidence, preserving context, designing geometry, controlling hardware, simulating behavior, or publishing what students learned.
Explore and remember
AI becomes useful when evidence survives
Microsoft Scout coordinates research, local tools, browser work, and deployment. GPT, Claude, and GitHub Copilot help us examine alternatives. GitHub preserves source history, while our LLMWiki connects evidence, decisions, procedures, and unresolved questions.
- Microsoft Scout
- GPT
- Claude
- GitHub Copilot
- GitHub
- LLMWiki
- MCP
Design across dimensions
Move from sketch to measured geometry
Onshape anchors collaborative CAD and official field evidence. FreeCAD supports open mechanical workflows. Blender turns CAD into visual, simulation, and game-ready assemblies. Affinity and Inkscape produce identity, diagrams, interface assets, and communication materials.
- Onshape
- FreeCAD
- Blender
- Affinity
- Inkscape
Build and control
Translate models into a physical robot
REV Robotics hardware and the FTC SDK run the robot. goBILDA components expand mechanical options. Limelight supports AprilTag vision. AMD Ryzen AI hardware gives local AI and simulation work a high-performance development platform.
- REV Robotics
- FTC SDK
- goBILDA
- Limelight
- AMD Ryzen AI
Connect through MCP
Let tools participate in the engineering loop
Model Context Protocol connections extend our workflow into Onshape, Blender, Roblox Studio, browser automation, and other controlled tools. The goal is not autonomous novelty; it is shorter distance between a question, an action, and verifiable evidence.
- Model Context Protocol
- Onshape MCP
- Blender MCP
- Roblox Studio MCP
Simulate, publish, and prove
Make work testable and accessible
Roblox Studio and Luau power the BIOBUZZ Roblox Simulation. Python, JavaScript, and Node.js build conversion and verification pipelines. Astro creates the public learning sites, and Cloudflare Pages delivers them over the team domains.
- Roblox Studio
- Luau
- Python
- JavaScript
- Node.js
- Astro
- Cloudflare Pages
BIOBUZZ Roblox Simulation
A strategy laboratory we can revisit
The OtterBots Roblox Simulation lets us explore field navigation, HIVE cycles, FLOWER timing, parking, alliance spacing, and driver decisions before every physical mechanism is ready.
It is an educational model, not proof of real robot performance. We use it to form hypotheses and plan physical tests, then record where the physical robot agrees or disagrees.
Microsoft Scout
Agentic research with proof
We use Microsoft Scout to connect research, local files, code, browser testing, cloud deployment, and team knowledge. A task is not complete because AI produced an answer; it is complete when the result is verified and persistent.
LLMWiki
A second brain for the team
Our LLMWiki stores source-backed findings, engineering decisions, procedures, and unresolved contradictions. It lets future students inherit context—not just files—and continue climbing from the current frontier.
Design and innovation
Useful beats complicated
Our goal is a robot whose mechanisms, wiring, software, controls, and maintenance access all support one understandable match strategy.
Integrated design
Evaluate the whole robot: packaging, protected wiring, service access, driver visibility, weight, and interaction between mechanisms.
Reliability first
A creative mechanism earns its place by working repeatedly. We track jams, missed detections, bounce-outs, reset behavior, and recovery time.
Human-centered controls
Driver controls should be predictable, clearly labeled, safely bounded, and practiced under realistic match pressure.
Professional development
Learn from the engineering community
Our learning plan is to deepen skills in mechanical design, control systems, computer vision, Roblox Simulation development, technical communication, and project leadership.
- Study the current Competition Manual, Team Updates, and official Q&A
- Use official FTC, REV, and vendor technical documentation
- Ask STEM professionals for technical feedback and career context
- Apply feedback in a prototype, program, test, or design revision
- Record what students learned and how it changed the work
Team continuity
Leave the next student a stronger starting point
Our public Code Lab, glossary, downloadable programs, diagrams, math explainers, and Roblox Simulation turn one season's work into reusable training material.
- Keep code and explanations together
- Document setup, assumptions, and safe test procedures
- Use shared vocabulary and named subsystem ownership
- Review milestones, risks, and open questions regularly
- Teach newer members by letting them operate and explain the system
Community impact
Share work people can actually use
We publish educational resources instead of reporting inflated audience numbers. When future outreach creates new teams, coaches, mentors, or volunteers, we will document those outcomes separately and accurately.
01
Open learning
Code Lab
Interactive FTC Blocks diagrams, full student-facing explanations, downloadable programs, and generated JavaScript.
Explore Code Lab →02Accessible language
Robot glossary
Plain-language definitions for the vision, math, programming, and control concepts used by the robot.
Open glossary →03Interactive strategy
Roblox Simulation
A public way to explore the BIOBUZZ field, practice decisions, and discuss strategy with shared visual context.
Launch Roblox Simulation →Student ownership and values
Every student should be able to explain the work
DiscoveryAsk questions and learn unfamiliar tools.
InnovationUse creativity and persistence to solve real problems.
ImpactTurn learning into resources others can use.
InclusionMake technical work understandable and invite every voice.
TeamworkShare ownership, review decisions, and help partners succeed.
FunCelebrate the experiments, progress, and people behind the robot.
Evidence boundary: This page describes current OtterBots work and operating goals visible through our public resources. It does not claim unverified competition results, recruitment outcomes, mentor relationships, or physical performance measurements.