Robot systems

1.In ROS 2 (Robot Operating System 2), a hospital delivery robot must ask its door controller to open a door and wait about 20 seconds while reporting progress, with the option to cancel. Which interface fits best?

2.What is the main difference between a ROS 2 topic and a ROS 2 service?

3.A teammate wants to send wheel velocity commands at 50 Hz by calling a ROS 2 service for each command. What is the best advice?

4.In ROS 2, what are parameters intended for?

5.A node in a parcel hub sorting cell calls a service synchronously from inside a subscription callback and the whole node freezes. It runs on a single-threaded executor. What is the most likely cause?

6.What does a ROS 2 callback group of the mutually exclusive type guarantee?

7.A lawn mower robot runs camera processing and a blade safety monitor as components in one process. Why might the team load them into one process rather than separate ones?

8.In ROS 2, what is a lifecycle (managed) node mainly used for?

9.A sidewalk delivery robot has two nodes publishing on /cmd_vel: teleoperation and autonomy. The wheels twitch between the two commands. What is the best fix?

10.Why do robot teams usually define custom ROS 2 message types in a separate interfaces package?

11.A map server publishes the map once at startup. A navigation node that starts 30 seconds later never receives it. Which quality of service (QoS) change fixes this?

12.A lidar driver publishes with best-effort reliability. A new obstacle node subscribes with reliable reliability and receives nothing. Why?

13.Why is best-effort reliability with a small history depth the usual choice for a 30 Hz camera stream?

14.An orchard drone sends a mission abort message that must not be lost on a lossy radio link. Which quality of service (QoS) settings fit best?

15.A subscriber uses keep-last history with depth 10 on a 100 Hz topic, and its callback stalls. Roughly how much recent data can its queue hold?

16.What does the ROS 2 deadline quality of service (QoS) policy let a subscriber detect?

17.A 10 Hz odometry topic should alert when it stalls. What deadline value would an engineer most likely set on it?

18.A pipeline-inspecting quadruped streams a 20 Hz pose topic over Wi-Fi to an operator laptop with reliable quality of service (QoS). When the link degrades, the operator view lags further and further behind. What is the best change?

19.A team keeps proportional-integral-derivative (PID) gains hard-coded in C++ and recompiles for every robot in a mixed fleet. What is the cleaner approach?

20.What is the main job of a ROS 2 launch file on a robot?

21.Two identical arms on one network both publish /joint_states, and their data mixes together. What is the standard fix?

22.Two robots on the same Wi-Fi network see each other's topics and one starts following the other's commands. Besides namespaces, what ROS 2 setting isolates them at the middleware level?

23.A parameter for a lab pipetting arm's maximum speed can be changed at runtime. What should the node do when someone sets it to a negative value?

24.A launch file starts a camera driver and a detector at the same time, and the detector crashes on startup because the camera is not ready. What is the most robust fix?

25.Under the common ROS frame convention, which node usually publishes the map to odom transform?

26.Why does a local obstacle-avoidance controller usually work in the odom frame rather than the map frame?

27.A warehouse robot's pose in RViz flickers between two positions about a meter apart, many times per second. Both an extended Kalman filter (EKF) node and the wheel driver are running. What is the most likely cause?

28.A robot is at x = 2 m, y = 0 m in the map, with yaw 90 degrees (facing +y). A lidar sits 0.5 m straight ahead of base_link. It sees an obstacle 1 m straight ahead of the lidar. Where is the obstacle in the map?

29.A robot is at x = 1 m, y = 1 m in the map with yaw 180 degrees. A point is at x = 2 m, y = 0 m in base_link. Where is the point in the map?

30.To express a camera detection in the map frame, which chain of transforms do you compose?

31.A transform lookup fails with an extrapolation into the future error when a node asks for the transform at a fresh lidar scan's timestamp. What is the usual fix?

32.A floor-cleaning robot draws lidar walls as curved smears whenever it spins in place, but walls look fine when it drives straight. What is the most likely cause?

33.A depth camera's points appear upside down and on the wrong side of the robot, though the depth values look right. What should you check first?

34.Under ROS conventions, how are the axes of a camera optical frame oriented?

35.On a kitchen robot arm, the wrist camera's frame stays frozen in RViz while the arm moves, but the base frames update. What is the most likely cause?

36.Why are fixed sensor mounting transforms usually published on /tf_static rather than /tf?

37.A robot's wheels have a radius of 0.1 m. The encoders report 2 full wheel revolutions with no slip. How far did the robot travel?

38.A team replays a bag file and every transform lookup fails, though live runs work. The bag was played with its clock option. What is the most likely missing setting?

39.What timestamp should a camera driver put in each image header?

40.A robot's timeout logic uses system wall-clock time. After the onboard computer syncs with a time server, several timeouts fire at once. What is the fix?

41.A sidewalk delivery robot fuses a lidar on one computer with a camera on another. The two clocks differ by 40 ms, and the robot moves at 2 m/s. Roughly how far apart will matched objects appear?

42.A robot has two onboard computers and needs their clocks aligned to well under a millisecond for sensor fusion. Which approach fits best?

43.A camera driver stamps images when the USB transfer completes, about 25 ms after exposure. Camera detections lag the lidar consistently. What is the best fix?

44.Why do nodes that combine a camera and a lidar often use an approximate time synchronizer rather than an exact one?

45.A robot without a battery-backed clock boots with its date set to 1970. Its software update agent fails to connect to the update server. What is the most likely cause?

46.An airport baggage tug drives at 1.5 m/s. Its obstacle data reaches the planner 150 ms after capture. How far has the tug moved since the data was captured?

47.A lidar spins at 10 Hz on a robot driving at 2 m/s. How far does the robot move during one full scan?

48.A perception pipeline on a 30 frames per second camera has three serial stages taking 12 ms, 15 ms, and 10 ms. What happens?

49.How can a node measure the end-to-end latency of camera images it receives?

50.A greenhouse picking arm sends full-resolution point clouds over Wi-Fi with best effort. About half never arrive, while small messages arrive fine. What is the most likely cause?

51.A motor controller receives velocity commands over a topic. What should it do with a command whose header stamp is 800 ms old when a fresh one is expected every 50 ms?

52.A topic rate tool shows a vineyard rover's camera at 30 Hz on the robot but only 12 Hz on the operator laptop. What does this suggest first?

53.A detection callback takes 80 ms, and it subscribes to a 20 Hz topic with keep-last depth 1. What will you observe?

54.A joint controller runs at 500 Hz. How much time does each cycle have, including reading sensors and writing commands?

55.What separates a hard real-time requirement from a soft real-time requirement?

56.A 1 kHz control loop on Linux usually finishes in 0.3 ms but sometimes takes 4 ms. Which code inside the loop is the most likely culprit?

57.A team's control loop has a low average cycle time but rare long spikes, and the arm sometimes jerks. Which metric should drive their fix?

58.A loop calls sleep(10 ms) after doing its work each cycle, aiming for 100 Hz. Over a minute it runs at about 92 Hz. What is the fix?

59.A mining tunnel crawler's wheel controller runs at 1 kHz, but its only feedback sensor updates at 50 Hz. What does the high loop rate buy?

60.Why do many robots run the motor current loop on a microcontroller and the planner on a Linux computer?

61.A team on a standard Linux kernel sees occasional 3 ms latency spikes in a high-priority control thread. What is a common step to reduce worst-case latency?

62.In a behavior tree, what does a sequence node do?

63.A hotel room service robot's behavior tree uses a fallback node with children: deliver via elevator, deliver via stairs lift, call staff for help. What does the fallback do?

64.A behavior tree action for drive to dock takes 40 seconds. How should the node behave when ticked?

65.A robot should abort its current task and go charge as soon as the battery drops below 15 percent, even mid-task. Which behavior tree pattern fits?

66.A flat state machine for a security patrol robot has grown to 40 states with transitions copied into each one for low battery and e-stop. What change helps most?

67.A library book robot retries a failed grasp forever, blocking the aisle for hours. What is the best design change?

68.Why should a robot's mission state machine have explicit fault states rather than only nominal ones?

69.A navigation stack works perfectly in simulation but oscillates on the real robot. Sensors and code are the same. What is the most likely gap?

70.A simulator reports a real-time factor of 0.5 during a test. What does that mean?

71.A team tunes a parcel-gripping routine in simulation and it fails often on real boxes. What is the most likely reason?

72.What is simulation most reliably good for in a robot software workflow?

73.A solar panel cleaning robot passes every simulation test, but on site its edge detector misses panel edges in low morning sun. What should the team add to the simulation?

74.Why should robot nodes read time through the ROS clock interface rather than calling the operating system clock directly?

75.A team wants to rerun a new perception algorithm on last week's field data. What must the bag contain?

76.A camera publishes 640 by 480 RGB images, 3 bytes per pixel, uncompressed at 30 Hz. About how much disk space does a 10 minute bag of this topic need?

77.A campus shuttle robot cannot record all sensors all day, but the team needs data around rare incidents. What is the common approach?

78.An engineer replays a bag on a development machine while the lidar driver node is also running there. The obstacle node behaves erratically. What is the most likely cause?

79.A rare bug appears in a recorded bag for a frame or two. What playback option helps most when stepping through it?

80.A team replays the same bag through their planner twice and gets slightly different paths. What is the most likely reason?

81.Why should a robot's text logs and its recorded sensor data share the same clock source?

82.What does hardware-in-the-loop (HIL) testing mean for a robot controller?

83.A team's simulation tests all pass on desktop machines, yet the robot misses control deadlines on its onboard computer. Which test stage would have caught this earlier?

84.Before a new motor driver runs on a full robot, what is a sensible first bench test?

85.A hardware-in-the-loop test of a pharmacy dispensing arm fails about once in 50 runs with no clear pattern. What should the team do first?

86.What is a sensible order of test stages for a change to a robot's navigation code?

87.Why should a robot's emergency stop (e-stop) cut motor power through a hardware circuit rather than rely only on a software message?

88.A motor controller stops the motors if it receives no command for longer than its watchdog timeout. A robot drives at 1.2 m/s with a 0.25 s timeout. How far can it travel after its last command before braking begins?

89.An operator releases the e-stop button on a recycling sorting arm. What should the arm do?

90.A watchdog on a lawn mower robot checks only that the navigation process is still alive. The planner thread hangs while the process stays up, and the mower keeps driving. What is the better design?

91.A security patrol robot loses its Wi-Fi link to the operator during autonomous patrol. What is a reasonable default behavior?

92.Where should a speed limit near people be enforced so a planner bug cannot exceed it?

93.What separates a controlled stop from an immediate power-cut stop on a heavy robot arm?

94.How often should a robot's e-stop chain be tested in the field?

95.A team will roll out new navigation software to 200 delivery robots using a 5 percent canary stage. How many robots get it first?

96.Why do many robot fleets use an A/B partition scheme for operating system updates?

97.When should a fleet of hospital delivery robots install a software update?

98.Two robots in the same fleet run the same software version but behave differently on the same route. What should the team check first?

99.After a fleet update, how should the team decide whether to continue the rollout?

100.Field robots on cellular links take hours to download each new full software image. What is a common improvement?

101.A robot fails in the field once a day in a way the team cannot reproduce at the office. What is the most useful first step?

102.A team writes a 1 kHz joint control loop in Python and sees periodic multi-millisecond stalls. What is the most likely cause?

103.Where does Python usually make good sense on a robot?

104.A Python node uses four threads to filter point clouds in pure Python loops, but uses only about one core. Why?

105.An onboard computer has 4 CPU cores. The robot's nodes together use 3.2 cores on average. What is the main concern?

106.A rail track inspection robot runs well for 20 minutes on hot days, then its perception rate halves. Indoors it never slows down. What is the most likely cause?

107.Two neural network models share one onboard GPU. The obstacle model occasionally misses its 50 ms budget when the other model runs. What is a sound first step?

108.A robot's debug logging prints every incoming lidar point at info level. The robot starts missing cycles. What is the best fix?