Perl, 110 characters

$p=144;$p+=(1,-17,-1,17)[($d+=($f[$p]^=2)+1)%4]for 1..<>;$f[$p]=1;print$_%17?'':$/,qw(_ @ #)[$f[$_]]for 0..288 

Number is read from the first line of STDIN. Rest of input is ignored.

Slightly more readable:

$p=144; $p += (1,-17,-1,17)[($d+=($f[$p]^=2)+1) % 4] for 1..<>; $f[$p]=1; print $_%17 ? '' : $/, qw(_ @ #)[$f[$_]] for 0..288 

Edits

• (112 → 111) No need to update $d with the modulo-4 value.

• (111 → 110) Can now inline the $d increment

Perl, 110 characters

$p=144;$p+=(1,-17,-1,17)[($d+=($f[$p]^=2)+1)%4]for 1..<>;$f[$p]=1;print$_%17?'':$/,qw(_ @ #)[$f[$_]]for 0..288 

Number is read from the first line of STDIN. Rest of input is ignored.

Slightly more readable:

$p=144; $p += (1,-17,-1,17)[($d+=($f[$p]^=2)+1) % 4] for 1..<>; $f[$p]=1; print $_%17 ? '' : $/, qw(_ @ #)[$f[$_]] for 0..288 

Edits

• (112 → 111) No need to update $d with the modulo-4 value.

• (111 → 110) Can now inline the $d increment

Addendum (109 characters)

We can have it one character shorter if you’re happy to have the special case of N=0 fail (it doesn’t output the @ character for the ant). All other inputs work correctly:

$p+=(1,-17,-1,17)[($d+=($f{$p+0}^=2)+1)%4]for 1..<>;$f{$p}=1;print$_%17-9?'':$/,qw(_ @ #)[$f{$_}]for-144..144 

The differences are that we now use %f instead of @f so we can use negative indices, and we iterate from -144..144 instead of 0..288. It saves having to initialise $p.

Perl, 112 characters

$p=144;$d+=($f[$p]^=2)+1,$p+=(1,-17,-1,17)[$d%=4]for 1..<>;$f[$p]=1;print$_%17?'':$/,qw(_ @ #)[$f[$_]]for 0..288 

Number is read from the first line of STDIN. Rest of input is ignored.

Slightly more readable:

$p=144; $d+=($f[$p]^=2)+1, $p+=(1,-17,-1,17)[$d%=4] for 1..<>; $f[$p]=1; print $_%17 ? '' : $/, qw(_ @ #)[$f[$_]] for 0..288 

Perl, 110 characters

$p=144;$p+=(1,-17,-1,17)[($d+=($f[$p]^=2)+1)%4]for 1..<>;$f[$p]=1;print$_%17?'':$/,qw(_ @ #)[$f[$_]]for 0..288 

Number is read from the first line of STDIN. Rest of input is ignored.

Slightly more readable:

$p=144; $p += (1,-17,-1,17)[($d+=($f[$p]^=2)+1) % 4] for 1..<>; $f[$p]=1; print $_%17 ? '' : $/, qw(_ @ #)[$f[$_]] for 0..288 

Edits

• (112 → 111) No need to update $d with the modulo-4 value.

• (111 → 110) Can now inline the $d increment

Perl, 112 characters

$p=144;$d+=($f[$p]^=2)+1,$p+=(1,-17,-1,17)[$d%=4]for 1..<>;$f[$p]=1;print$_%17?'':$/,qw(_ @ #)[$f[$_]]for 0..288 

Number is read from the first line of STDIN. Rest of input is ignored.

Perl, 112 characters

$p=144;$d+=($f[$p]^=2)+1,$p+=(1,-17,-1,17)[$d%=4]for 1..<>;$f[$p]=1;print$_%17?'':$/,qw(_ @ #)[$f[$_]]for 0..288 

Number is read from the first line of STDIN. Rest of input is ignored.

Slightly more readable:

$p=144; $d+=($f[$p]^=2)+1, $p+=(1,-17,-1,17)[$d%=4] for 1..<>; $f[$p]=1; print $_%17 ? '' : $/, qw(_ @ #)[$f[$_]] for 0..288