Skip to main content
. Author manuscript; available in PMC: 2009 Sep 22.
Published in final edited form as: J Artif Intell Res. 2008 Jul 1;32(2):663–704.

Table 2.

Comparison of different search heuristics on the Tag environment using the Blind policy as a lower bound.

Heuristic Return EBR (%) LBI Belief Nodes Nodes Reused (%) Online Time (ms)
RTBSS(5) −10.31 ± 0.22 22.3 ± 0.4 3.03 ± 0.07 45066 ± 701 0 580 ± 9
Satia and Lave −8.35 ± 0.18 22.9 ± 0.2 2.47 ± 0.04 36908 ± 209 10.0 ± 0.2 856 ± 4
AEMS1 −6.73 ± 0.15 49.0 ± 0.3 3.92 ± 0.03 43693 ± 314 25.1 ± 0.3 814± 4
HSVI-BFS −6.22 ± 0.19 75.7 ± 0.4 7.69 ± 0.06 64870 ± 947 54.1 ± 0.7 673 ± 5
BI-POMDP −6.22 ± 0.15 76.2 ± 0.5 7.81 ± 0.06 79508 ± 1000 54.6 ± 0.6 622 ± 4
AEMS2 −6.19 ± 0.15 76.3 ± 0.5 7.81 ± 0.06 80250 ± 1018 54.8 ± 0.6 623 ± 4