AYX / KNOWLEDGE BASE
Good questions. Clear answers.
Practical answers, sources and examples across robotics, AI, smart homes and everyday technology.
DISCOVER / SEARCH
Search the knowledge library
Type a term or question to search across every subject.
INDEX / 01–10
Explore by subject
Industrial robotics
Robot fundamentals, selection and integration.
Robot buying & ownership
New, used and refurbished robots, maintenance and ROI.
Education & ROS 2
Learn robotics, programming and simulation.
Mobile robots & navigation
AMRs, AGVs, LiDAR and autonomous navigation.
Machine vision
Vision, inspection, sensing and robot guidance.
Humanoids & quadrupeds
Real-world uses and practical limits of legged robots.
AI, LLMs & agents
Algorithms, language models and AI agents.
Physical AI & robot models
Embodied AI, NVIDIA, VLA and decision models.
Home labs, smart home & privacy
Run, secure and connect technology at home.
Everyday technology & learning
Devices, everyday skills, family life and science.
START / QUESTIONS
Essential questions
- What is the difference between a robot, robotics and automation?
- Should I buy a new, used or refurbished industrial robot?
- Why does a complete robot application cost more than the robot arm?
- How do payload, centre of gravity and reach affect robot selection?
- How do I choose between articulated, SCARA, delta and Cartesian robots?
- How do I calculate robot payback without exaggerating savings?
- What is the difference between an AMR and an AGV?
- What is machine vision, and when does it need AI?
- What is ROS 2, and is it a robot operating system in the usual sense?
REFERENCE / A–Z
A–Z glossary
- Accuracy / repeatability
- Closeness to a intended target versus consistency of repeated results.
- AGV
- Automated guided vehicle; not “AVG”. Navigation method must be checked rather than inferred from the abbreviation.
- AMR
- Autonomous mobile robot; autonomy does not remove site, integration or safety constraints.
- Axis / degree of freedom
- Related motion/kinematic concepts; not always interchangeable counts.
- CADR
- Clean air delivery rate; assess the relevant pollutant, room and usable operating setting.
- Digital twin
- A maintained model linked to relevant real-system data/behaviour, not merely any 3D drawing.
- FAT / SAT
- Factory acceptance test / site acceptance test; agreed evidence at different stages.
- Imitation learning / reinforcement learning
- Learning from demonstrations / learning through an objective and feedback.
- IMU
- Inertial measurement unit used with other sources to estimate motion.
- JEPA
- Joint-embedding predictive architecture; unrelated to the product spelling Jev.
- Jev / Laya
- Distinct decision-model research subjects, not corrections of JEPA or VLA.
- LiDAR
- Light-based ranging; navigation capability does not establish safety certification.
- LLM / token / context
- Language model / model text unit / available input-output context boundary.
- Matter / Thread
- Interoperability/application ecosystem versus an IP-based mesh networking technology.
- Nebula / exoplanet
- Gas/dust cloud / planet beyond the Solar System; neither term establishes life.
- Open-weight
- Accessible model weights; not necessarily unrestricted or fully open-source.
- Passkey
- Public-key authentication designed for phishing resistance; recovery still matters.
- Payload / centre of gravity / inertia
- Different load properties; tool and workpiece both matter.
- RAG
- Retrieval-augmented generation; retrieved context still needs relevance and trust checks.
- ROS 2
- Robotics middleware and ecosystem; not by itself a complete real-time operating system or safety controller.
- Segmentation / detection
- Region/pixel assignment versus locating objects; select for the task.
- Sim-to-real
- The challenge of transferring behaviour from simulation to physical conditions.
- SLAM
- Simultaneous localisation and mapping; separate from planning and motion control.
- TCO / simple payback
- Lifecycle cost / illustrative investment-to-net-benefit ratio; different economic measures.
- TCP
- Tool centre point used to describe tooling position and motion.
- Training / inference
- Learning/adapting model parameters versus running a model on an input.
- VLM / VLA
- Vision-language model / vision-language-action model; different expected outputs.