Information processing method, task execution method, apparatus, device and medium

An electronic device, comprising:

• a processor;
• and
• a memory, configured to store executable instructions of the processor;
• wherein the processor is configured to perform, by executing the executable instructions, acts comprising:
• processing, through a target visual encoding model in a target analysis model, obtained image information to be analyzed, to obtain a corresponding target sequence, the image information to be analyzed comprising environment image information of an environment in which an action execution apparatus is located and image information of the action execution apparatus;
• fusing, through a target feature fusion model in the target analysis model, the target sequence and obtained text information to be analyzed, to obtain a target fusion result, the text information to be analyzed corresponding to the image information to be analyzed;
• processing the target fusion result through a target task analysis model in the target analysis model to obtain target task information;
• controlling the action execution apparatus to perform an action corresponding to the target task information;
• and
• wherein the target analysis model is obtained by training an initial analysis model;
• the initial analysis model comprises an initial visual encoding model and an initial feature fusion model;
• during the initial analysis model being trained to be the target analysis model, the initial visual encoding model is trained to be the target visual encoding model, and the initial feature fusion model is trained to be the target feature fusion model;
• and at least one of the initial visual encoding model or the initial feature fusion model is obtained by pre-training.

A non-transitory computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, causes the processor to perform acts comprising:

• processing, through a target visual encoding model in a target analysis model, obtained image information to be analyzed, to obtain a corresponding target sequence, the image information to be analyzed comprising environment image information of an environment in which an action execution apparatus is located and image information of the action execution apparatus;
• fusing, through a target feature fusion model in the target analysis model, the target sequence and obtained text information to be analyzed, to obtain a target fusion result, the text information to be analyzed corresponding to the image information to be analyzed;
• processing the target fusion result through a target task analysis model in the target analysis model to obtain target task information;
• controlling the action execution apparatus to perform an action corresponding to the target task information;
• and
• wherein the target analysis model is obtained by training an initial analysis model;
• the initial analysis model comprises an initial visual encoding model and an initial feature fusion model;
• during the initial analysis model being trained to be the target analysis model, the initial visual encoding model is trained to be the target visual encoding model, and the initial feature fusion model is trained to be the target feature fusion model;
• and at least one of the initial visual encoding model or the initial feature fusion model is obtained by pre-training.