https://toorova.com/articles In-depth, sourced explainers on how AI and robots work. en-us Tue, 03 Nov 2026 12:00:00 +0000 https://toorova.com/articles/how-does-a-robot-know-where-it-is https://toorova.com/articles/how-does-a-robot-know-where-it-is Tue, 03 Nov 2026 12:00:00 +0000 Robots and robotics A robot estimates where it is by combining two kinds of information. Relative sensors, like wheel encoders and an IMU, track how far it has moved since it last knew its position. They're fast and smooth, but their errors pile up. Absolute references, like GPS, known landmarks or a map matched against a lidar or camera, don't drift but are noisy or patchy. A filter, often a Kalman filter, blends the two, trusting each in proportion to how reliable it is. https://toorova.com/articles/how-do-large-language-models-work https://toorova.com/articles/how-do-large-language-models-work Tue, 03 Nov 2026 12:00:00 +0000 How AI works A large language model, like the one behind ChatGPT, is a program that predicts the next token (a word or piece of a word) given the text so far. It turns that text into a list of probabilities for every possible next token, picks one, adds it to the text and repeats. It learned those probabilities by practicing next-token prediction on hundreds of billions of tokens of text, or more. It doesn't look facts up unless it's connected to a search tool.