Practice
Real, interview-grade questions — anonymized & vetted. Subjective answers get AI feedback.
62 questions
- hardWhy is the Kalman filter optimal — and when does it stop being optimal?Kalman / EKF / UKF· Subjective · AI-graded· Practice question
- mediumBehavior of the Kalman gainKalman / EKF / UKF· Multiple choice· General
- easyEliminating steady-state errorControl Theory (PID / LQR / MPC)· Multiple choice· General
- easyEKF prediction covariance dimensionKalman / EKF / UKF· Numerical· General
- mediumWhy loop closure mattersSLAM· Subjective · AI-graded· Practice question
- mediumAdmissible heuristics in A*Motion Planning· Multiple choice· General
- easyEigenvalues and Eigenvectors of a Rotation MatrixLinear Algebra· Subjective · AI-graded· General
- easyDeterminant of a Valid Rotation MatrixLinear Algebra· Numerical· General
- easyPosterior probability of an obstacle from a noisy proximity sensorProbability & Statistics· Numerical· General
- easyExplain the bias-variance tradeoff for an estimatorProbability & Statistics· Subjective · AI-graded· General
- mediumGradient Descent vs. Gauss-Newton for Robotics Least-SquaresCalculus & Optimization· Subjective · AI-graded· General
- mediumConvergence Behavior at a Saddle Point of the Cost SurfaceCalculus & Optimization· Multiple choice· General
- mediumWhy a 4x4 homogeneous transform instead of separate R and t?Rigid-Body Motion & Transforms· Subjective · AI-graded· General
- mediumNet rotation angle of two composed 90-degree rotationsRigid-Body Motion & Transforms· Numerical· General
- mediumCamera Calibration and the Pinhole ModelComputer Vision· Numerical· General
- mediumRobustness of Vision Perception to Real-World ConditionsComputer Vision· Subjective · AI-graded· General
- mediumComputing F1 from a confusion matrixML Fundamentals· Numerical· General
- mediumDiagnosing and addressing the bias-variance tradeoffML Fundamentals· Subjective · AI-graded· General
- mediumSingularities in Inverse Kinematics: Detection and HandlingKinematics (FK / IK)· Subjective · AI-graded· General
- mediumForward Kinematics of a Planar 2-Link ArmKinematics (FK / IK)· Numerical· General
- mediumProcess vs. Measurement Noise in Kalman Filter TuningBayesian Filtering· Subjective · AI-graded· General
- mediumDiscrete Bayes Filter: One Measurement UpdateBayesian Filtering· Numerical· General
- hardDiagnosing and fixing exploding gradients in a deep recurrent/Transformer training runDeep Learning (PyTorch)· Subjective · AI-graded· General
- hardCounting parameters in a 2D convolution layer (with bias)Deep Learning (PyTorch)· Numerical· General
- hardSkew-symmetry of the manipulator dynamics and the passivity propertyDynamics· Subjective · AI-graded· General
- hardReflected inertia through a gear reductionDynamics· Numerical· General
- hardDeadly Triad and Off-Policy Value DivergenceReinforcement Learning· Subjective · AI-graded· General
- hardGAE: Effect of the lambda Parameter on the Bias-Variance TradeoffReinforcement Learning· Multiple choice· General
- hardScalar LQR: solving the Riccati equation for the optimal gainControl Theory (PID / LQR / MPC)· Numerical· General
- hardMPC vs LQR: when the extra machinery of MPC earns its costControl Theory (PID / LQR / MPC)· Subjective · AI-graded· General
- hardParticle Deprivation and the Cost of ResamplingParticle Filters· Subjective · AI-graded· General
- hardComputing Effective Sample SizeParticle Filters· Numerical· General
- hardEKF Consistency, Linearization Error, and Choosing Between EKF and UKFSensor Fusion· Subjective · AI-graded· General
- hardStatic 1D Fusion: Optimal Inverse-Variance Weighting and Fused VarianceSensor Fusion· Numerical· General
- mediumSampling-based planners: RRT vs. RRT* and the role of optimalityMotion Planning· Subjective · AI-graded· General
- mediumHeuristic admissibility and A* optimalityMotion Planning· Multiple choice· General
- mediumChoosing and tuning DDS QoS for a ROS 2 sensor streamROS / ROS2· Subjective · AI-graded· General
- mediumSizing a ROS 2 subscriber queue from publish rate and processing timeROS / ROS2· Numerical· General
- hardConsistency and the Observability of EKF-SLAMSLAM· Subjective · AI-graded· General
- hardInformation-Form Fusion of Two Range MeasurementsSLAM· Numerical· General
- mediumDesigning the FBD and EOM for a friction-driven robot wheelClassical Mechanics· Subjective · AI-graded· General
- easyAngular speed after a robot arm retracts a mass (conservation of angular momentum)Classical Mechanics· Numerical· General
- mediumChoosing a sample rate and anti-aliasing for an IMUSignal Processing· Subjective · AI-graded· General
- easyAliased frequency from undersamplingSignal Processing· Numerical· General
- mediumWhy an IMU alone cannot be trusted for longSensors & How They Work· Subjective · AI-graded· General
- easyHeading error from a constant gyro biasSensors & How They Work· Numerical· General
- mediumSizing a motor and gear ratio for a robot jointActuators & Motors· Subjective · AI-graded· General
- easyOutput torque of a geared DC motor at a given currentActuators & Motors· Numerical· General
- mediumPriority inversion in a real-time control loopEmbedded & Real-Time· Subjective · AI-graded· General
- easyUART throughput sizingEmbedded & Real-Time· Numerical· General
- mediumWhy co-train a VLA on web data instead of fine-tuning only on robot dataFoundation Models· Subjective · AI-graded· General
- mediumAction tokenization vs. flow-matching action heads in VLAsFoundation Models· Multiple choice· General
- mediumWhy behavior cloning compounds errors and how DAgger helpsImitation Learning· Subjective · AI-graded· General
- mediumChoosing a policy class for multimodal demonstrationsImitation Learning· Multiple choice· General
- hardToken binning vs. continuous action decoding in VLAsVision-Language-Action (VLA)· Subjective · AI-graded· General
- mediumWhy internet pretraining helps VLA generalizationVision-Language-Action (VLA)· Multiple choice· General
- mediumWhy latent imagination is sample-efficientWorld Models & Sim-to-Real· Subjective · AI-graded· General
- mediumWhat domain randomization fundamentally trades offWorld Models & Sim-to-Real· Multiple choice· General
- mediumDesigning a multi-robot training data formatData & ML Infrastructure· Subjective · AI-graded· General
- mediumWhy GPU-parallel simulation changes RL training economicsData & ML Infrastructure· Multiple choice· General
- mediumWhy a reliable subscriber misses a latched topicMiddleware & System Architecture· Subjective · AI-graded· General
- easyWhat the RMW layer is forMiddleware & System Architecture· Multiple choice· General