Algorithm: An algorithm for extracting operation patterns of industrial robots

Input: Hash sequence 0: $Q = H_{1} H_{2} \dots H_{n}$

Output: Hash sequence 3: $Q^{3} = H_{1}^{1} H_{2}^{1} \dots H_{i}^{1}$

Definition 1: Definition $H_{i}$ , $i = 1 \dots n$ is the image hash value calculated after image segmentation of a single frame industrial robot.

Definition 2: Definition $H_{i}^{1} ≅ H_{i}^{2}$ robot attitude vector $H_{i}^{1}$ and $H_{i}^{2}$ approximately equal, so that $D (H_{1}, H_{2}) \leq δ$ , where D(H₁, H₂) is Hamming distance between H₁ and H₂, and δ is the given similarity threshold.

Steps:

1) Set Q suffix sequence of $S (Q) = {H_{i} H_{i + 1} \dots H_{n} | i = 1 \dots n}$

2) Make any hash sequence $Q^{1}$ , $Q^{2} \in S (Q)$

3) st. $Q^{1} + Q^{2} = S (Q)$

4) Let $Q^{1} = H_{1}^{1} H_{2}^{1} \dots H_{n}^{1}$ and $Q^{2} = H_{1}^{2} H_{2}^{2} \dots H_{n}^{2}$

5) get the common prefix subsequence of Q¹, Q²

6) for i in n:

7) if $H_{1}^{1} ≅ H_{1}^{2} \land H_{2}^{1} ≅ H_{2}^{2} \land \dots \land H_{i}^{1} ≅ H_{i}^{2},$

8) Add to common prefix subsequence set H(Q)

9) end for

10) Q³ = Softmax(H(Q))

11) Output Q³