Robotics

Robotics is the discipline of building machines that sense their surroundings, decide what to do, and move through the physical world to do it. Underneath every robot — a factory arm, a warehouse rover, a delivery drone, a search-and-rescue hexapod — sits the same stack of problems: work out where your body actually is (kinematics), work out where it needs to go (planning), work out how to drive the motors there smoothly (control), and work out what the sensors are actually telling you about the world (perception). This hub gathers every interactive robotics simulation on mysimulator.uk into one guided starting point, so instead of reading a robotics textbook cover to cover you can drag a target, watch a chain of joints reach for it, and see exactly which numbers the algorithm is computing on every frame.

The foundation of the topic is kinematics: the geometry that relates a robot's joint angles to the position of its end-effector. The robot arm kinematics simulation lets you click anywhere on screen and watch a 3-joint arm compute the inverse problem live using the Jacobian transpose method, gracefully handling the singularities and joint limits that make inverse kinematics one of the classic hard problems in robotics. The inverse kinematics (FABRIK) simulation solves the same underlying problem with a different, faster technique — Forward And Backward Reaching Inverse Kinematics converges in a handful of iterations by alternating passes from the target back to the base and back again, which is why it is the method of choice in real-time character animation and increasingly in robotics itself, including branching, spider-like chains with multiple end-effectors.

Knowing the target joint angles is only half the job — something has to actually drive the motors there without overshooting, oscillating, or drifting off target under load. The PID controller simulation demonstrates the single most widely deployed feedback algorithm in engineering: proportional, integral and derivative terms combine into one control signal, u = Kp·e + Ki·∫e + Kd·ė, and the same three-term loop that keeps a quadcopter level in a gust of wind also holds a thermostat, a cruise control system and countless industrial processes at their setpoint. The hexapod walker simulation shows what happens when you stack kinematics and control together at scale: each of six legs solves its own 2-link inverse kinematics analytically via the law of cosines, and a gait planner — tripod, alternating two triplets of legs, or wave, lifting one leg at a time — coordinates all six into stable, continuous locomotion across flat or bumpy terrain.

The drone swarm simulation moves from a single robot's joints to the coordination of many independent agents at once. Built on Boids-style separation, alignment and cohesion forces, the swarm holds a flock, circle, grid or V-shape formation with no leader and no central controller — each drone only reacts to its near neighbours, and the formation emerges from those local rules exactly the way real multi-robot systems and biological flocks do. The LiDAR scanning simulation closes the loop with perception: a rotating 2D sensor fires rays that ray-march out to the nearest wall, and the returns are assembled into both a noisy point cloud and a log-odds occupancy grid — free space along each ray, occupied at the point it hits something — which is exactly the raw sensor data that real SLAM (simultaneous localization and mapping) systems consume to let a robot build a map of an unknown space while tracking its own position inside it.

Together these six simulations cover the four pillars every working robot needs: kinematics to know where its body is, planning and gait coordination to decide how to move it, feedback control to actually execute that motion accurately, and perception to sense the world it is moving through. Follow the learning path below for a suggested order that builds from a single joint to a full sensing, coordinated system, browse the full grid for anything that catches your eye, or jump straight to the Robotics & Kinematics and Engineering category pages for the complete lists.

What makes these simulations different from a diagram in a textbook is that every one of them is a real numerical solver running live in your browser, not a pre-rendered animation. The robot arm and FABRIK simulations genuinely recompute the inverse kinematics solution every frame as you drag the target, so a singular pose or an out-of-reach target behaves exactly the way a real controller would have to handle it — clamping joint limits, damping near singularities, converging iteration by iteration. The PID simulation genuinely integrates the plant's equations of motion and feeds back the Kp, Ki and Kd terms you set, so an over-aggressive derivative gain really does make the system ring, and an under-tuned integral term really does leave a lasting steady-state error, precisely as the mathematics predicts. That distinction matters for engineering students preparing for a controls exam, for hobbyists building a real hexapod or quadcopter who want intuition before they solder anything, and for anyone who wants a feel for robotics that survives contact with the real equations — because the numbers on screen are the numbers the mathematics actually produces, not an artist's impression of them.