Impossible List

Crossed-out tasks are completed milestones; open tasks are future goals. Original dates are preserved. Dates marked "approx." are estimates for milestones without a recorded completion date.

Drone

• Design and build the custom frame and STM32 attitude-control prototype approx. 2024–2025
• First remote-controlled flight 10/10/2025
• Compare PID, LQR and constrained MPC with a navigation EKF in MATLAB approx. 2026 · simulation only
• Fix complementary-filter gyro mapping and add axis-isolation regression tests 06/10/2026
• Validate attitude estimation and controller tuning with repeatable flight logs
• Integrate GNSS RTK position estimation on hardware
• Demonstrate autonomous flight with staged safety and failsafe testing

6 DOF Arm

• Implement inverse kinematics 10/08/2023
• Historical ROS 2 / micro-ROS prototype milestone 28/09/2023 — consumer not in current public source
• Create the operator GUI 05/10/2023
• Install joint encoders 13/10/2023
• Implement encoder-based teach recording and replay 23/10/2023
• Explore encoder-feedback control 06/01/2024 — not a published six-axis servo
• Correct target rotations and ROS radians; add FK–IK round-trip tests 06/10/2026
• Publish and test an end-to-end ROS 2 hardware command bridge
• Migrate the embedded controller to STM32H7
• Develop continuous six-axis encoder-feedback control and measure repeatability

Balancing Robot

• Build the first balancing robot 28/01/2023
• Upgrade the MCU 05/11/2023
• Redesign the hardware 10/11/2023
• Implement LQR control 04/05/2025
• Integrate wheel-encoder and balance-offset pitch Kalman filters for stable LQR balancing approx. 2026
• Demonstrate position hold, trajectory tracking and keyboard teleoperation approx. 2026
• Publish timestamped disturbance-rejection logs and measure control-loop jitter
• Evaluate MPC in simulation before considering an embedded implementation

MicroDuck

• Complete mechanical assembly, actuator bring-up and joint calibration
• Set up the robot model and reinforcement-learning training environment
• Train and evaluate a walking policy in simulation
• Transfer the policy to hardware with controlled sim-to-real tests
• Publish a walking demonstration and reproducible setup

MicroDuck is in the build phase. Policy training and hardware walking are not completed milestones.

Other builds

• Create this website 01/08/2023
• Assemble the first tiny vario PCB and enclosure April 2026
• Revise the tiny vario PCB layout June 2026
• Develop a torque/position-sensing actuator and validate its sensing electronics
• Explore delay-compensated bilateral teleoperation with the actuator