# Robot

Here’s my robot. The coolest part is the predictive tracker, which
makes him a pretty good shot.

Kinda long for a quiz submission, I know. But it’s a tough problem.

regards,
Ed

require ‘robot’
require ‘matrix’

BOT_MAX_SPEED = 8
BULLET_SPEED = 30

class NotEnoughData < RuntimeError; end

class PredictiveTracker
def initialize(size = 4)
@size = size || 4

``````@x = Array.new(@size,0)
@y = Array.new(@size,0)
@t = Array.new(@size,0)

@most_recent = 0
@solution = nil
``````

end

def mark(x,y,time)
@most_recent = (@most_recent + 1) % @size
@x[@most_recent] = x
@y[@most_recent] = y
@t[@most_recent] = time

``````# Update solution, if possible
@solution = solve
``````

rescue
end

def solve
xoff, vx = linfit(@t, @x)
yoff, vy = linfit(@t, @y)
[vx,vy,xoff,yoff]
end

# predicts target location at the given time

def predict(time)
raise NotEnoughData unless @solution
vx, vy, xoff, yoff = @solution
[xoff + vxtime, yoff + vytime]
end

def aim_point(my_x, my_y, time)
raise NotEnoughData unless @solution
t = 0
loop do
x,y = predict(time+t)
break if vs(vd([x,y],[my_x,my_y])) < BULLET_SPEEDBULLET_SPEEDt*t
t += 1
raise NotEnoughData if t > 100
end
predict(time+t)
end

def firing_angle(my_x,my_y,time)
x,y = aim_point(my_x,my_y,time)
Math.atan2(my_y-y,x-my_x).to_deg
end

# returns [a,b] such that a + bx = y is least squares linear fit

def linfit(x,y)
sum_x = 0.0
sum_y = 0.0
sum_prod = 0.0
sum_x2 = 0.0
n = x.size
n.times {|i|
sum_x += x[i]
sum_y += y[i]
sum_prod += x[i]y[i]
sum_x2 += x[i]x[i]
}
b = (n
sum_prod - sum_x
sum_y) / (nsum_x2 - sum_xsum_x)
a = (sum_y - b*sum_x) / n
raise NotEnoughData if a.nan? or b.nan?
[a,b]
end

# Vector difference

def vd(a,b)
a.zip(b).map{|a,b| a - b}
end

def vs(a)
dp(a,a)
end

# Dot product

def dp(a,b)
a.zip(b).map{|a,b| a*b}.inject(0){|a,b| a+b}
end

end

class EdBot
include Robot

def tick events
startup if time == 0
update_gun

``````accelerate 1

turn_gun(@gun_velocity - @angular_velocity)
turn(@angular_velocity)
``````

end

if events[‘robot_scanned’].empty?
if @saw_target
high_low
else
low_low
end
@saw_target = false
else
td = events[‘robot_scanned’].min.first
if @saw_target
high_high(td)
else
low_high(td)
end
@saw_target = true
end
end

def low_low

``````if @downticks > 0
@downticks -= 1
if @downticks == 0
end
end
``````

end

def low_high(dist)
@uptick_dist = dist
end

def high_high(dist)
@uptick_dist = dist
end

def high_low
@downticks = 8
end

def beam_center
end

fire 3
end
end

def update_gun
@gun_velocity = clamp(diff,-30,30)
trigger(diff)
rescue NotEnoughData
end

def wall_force(range)
(2**((battlefield_width -
range)/50.0))/(2**(battlefield_width/50.0))
end

ranges = [x-size, battlefield_height-size-y,
battlefield_width-size-x, y-size]
normals = [0, 90, 180, 270]
forces = ranges.map {|r| wall_force®}

``````@xforce = forces[0] - forces[2]
@yforce = forces[3] - forces[1]
fa = Math.atan2(-@yforce,@xforce).to_deg

goal = (goal + 180) % 360
end

diff = angle_direction(goal, fa)
goal += diff*(forces.max)
``````

end

def startup
@saw_target = false

``````@target_x = @target_y = 0

@downticks = 0

@gun_velocity = 0
@angular_velocity = 0

@tracker = PredictiveTracker.new

@log = File.open("edbot.log","a")
@log.write "Starting up!\n"
``````

end

def angle_difference(a,b)
d = (a % 360 - b % 360).abs
d > 180 ? 360 - d : d
end

# To turn from a toward b, how should you turn?

def angle_direction(a,b)
magnitude = angle_difference(a,b)
if angle_difference(a + 1, b) < magnitude
magnitude
else
-magnitude
end
end

end

def angle_average(a,b)
(angle_direction(a,b) / 2 + a) % 360
end

# How much can the angle to the target change in one tick?

def target_angular_speed
360 * BOT_MAX_SPEED / (2 * Math::PI * target_distance)
end

Math.atan2(y- @target_y, @target_x - x).to_deg
end

def target_distance
Math.sqrt((@target_x - x)**2 + (@target_y - y)**2)
end

@target_y = y - distance * Math.sin(rads)
@target_x = x + distance * Math.cos(rads)
@tracker.mark(@target_x,@target_y,time)
@log.write("#{@target_x}\t#{@target_y}\t#{time}\n")
end

def clamp(var, min, max)
val = 0 + var # to guard against poisoned vars
if val > max
max
elsif val < min
min
else
val
end
end

end

I guess since filenames are significant I should have attached it
instead of inserting it. Here it is again.

regards,
Ed