Answers

U11.7 Project: find the one that is green

Vision · University · about 50 min

What this lesson is about

Three identical shapes, one job, and a task the depth sensor cannot begin.

• The job
• What is worth thinking about
• A shape to start from
• What comes next

Questions 6 marks in all

1. [1 mark]Three identical balls differ only in colour. Why can the depth sensor not pick out the green one?

   1. AIt measures a property of space, and identity lives in the surfaces, which only the camera sees
   2. BIts range is too short to reach the balls
   3. CIts readings are too noisy
   4. DIt cannot measure bearings
   Answer: A. To the depth sensor they are three identical bumps. No cleverness with range data recovers a property it never measured.

2. [1 mark]The survey finds three 6 cm balls. What does this print?

   import math
   F, R = 92.4, 3.0
   seen = {"red": (100, 14), "green": (200, 28), "blue": (250, 9)}
   for colour, (cx, width) in seen.items():
       got = 2 * R * F / width, math.degrees(math.atan((cx - 160) / F))
       print(colour, "%.0f cm at %.0f degrees" % got)

   Answer:

   red 40 cm at -33 degrees
   green 20 cm at 23 degrees
   blue 62 cm at 44 degrees

   Range is 554.4 / width: 39.6, 19.8 and 61.6 cm. Bearings are atan((cx - 160) / 92.4): -33, 23 and 44 degrees.

3. [1 mark]Why turn to the green ball's bearing and servo on its apparent size, rather than work out its coordinates and drive there open loop?

   1. AThe range is far worse than the bearing, and closing the loop on size lets the measurement improve as the robot gets closer
   2. BCoordinates cannot be computed from a bearing and a range
   3. COpen loop driving is not allowed on the mat
   4. DServoing is faster than driving a straight line
   Answer: A. Range from size degrades as d squared, so the survey range is the weakest number you have. Near the ball it is a precision instrument.

4. [1 mark]Why must the survey be a separate phase before the approach?

   1. AThe camera runs one detector at a time, so it cannot watch for red while servoing on green
   2. BThe balls move once the robot sets off
   3. CThe depth sensor must be switched off while the camera runs
   4. DThe bearings change units when the robot moves
   Answer: A. A small camera pipeline has one frame buffer and one configuration. Switching detectors costs time, so plan the survey and commit.

5. [1 mark]The approach should stop 20 cm from a 6 cm ball. About how many pixels wide should it look, to 1 decimal place, with f = 92.4?

   Answer: 27.7 (accept within 0.4). w = 92.4 x 6 / 20 = 27.7 pixels, about 28.

6. [1 mark]Which choices help the robot stop short of the ball without touching it?

   Tick every answer that is true.

   1. AUse the dead band trick rather than a large forward gain
   2. BStop on a band of widths rather than one exact width
   3. CRaise the forward gain so it settles before it can overshoot
   4. DPlan the approach from the survey range and drive it without looking
   Answer: A, B. Overshoot comes from a high gain in a slow loop. A band stops the loop hunting, and driving blind on a poor range is how balls get hit.

The task: find the one that is green

from bugbot import *
import math
connect()

F, R = 92.4, 3.0
TARGET = "green"

from bugbot import *
import math
connect()

F = 92.4
R = 3.0                              # every ball is 6 cm across
STANDOFF = 22.0
WANT = 2 * R * F / STANDOFF
TARGET = "green"

def survey(colour):
    """Range and bearing of the largest blob of one colour, or None."""
    set_cv("blob", colour)
    wait(0.3)
    seen = blobs()
    if not seen:
        return None
    cx, cy, area, x0, y0, x1, y1, aspect = seen[0]
    return 2 * R * F / (x1 - x0), math.degrees(math.atan((cx - 160) / F))

for colour in ("red", "green", "blue"):
    got = survey(colour)
    if got is None:
        print(colour + ": not in view")
    else:
        print("%s: %.1f cm, bearing %.1f" % (colour, got[0], got[1]))

set_cv("blob", TARGET)
wait(0.2)
last = 1.0
for tick in range(700):
    seen = blobs()
    if not seen:
        drive(0, 0, 25 * last)
        wait(0.1)
        continue
    cx, cy, area, x0, y0, x1, y1, aspect = seen[0]
    w = x1 - x0
    err = cx - 160

    last = 1.0 if err > 0 else -1.0
    rot = max(-55, min(55, 0.35 * err))
    if abs(err) < 6:
        rot = 0
    elif abs(rot) < 17:
        rot = 17 * (1 if rot > 0 else -1)

    gap = WANT - w
    fwd = max(-40, min(45, 4.0 * gap))
    if abs(gap) < 1.5:
        fwd = 0
    elif abs(fwd) < 18:
        fwd = 18 * (1 if fwd > 0 else -1)
    if abs(err) > 60:
        fwd = 0

    drive(fwd, 0, rot)
    plot("cx", cx)
    plot("width", w)
    wait(0.1)
    if fwd == 0 and rot == 0:          # both errors inside their bands: arrived
        break
stop()
wait(0.5)
print("arrived at the", TARGET, "ball")