Gesture-based systems for human following

A system comprising:

• a vehicle;
• an image sensor connected to the vehicle;
• and
• a processing apparatus configured to:
• detect a human in a first image captured using the image sensor;
• input at least a portion of the first image to a first machine learning model to obtain a first pose of the human, wherein the first pose includes a position of a wrist, an elbow, a shoulder, a waist, a knee, an ankle, or a knuckle;
• compare the first pose to pose parameters for an authentication gesture, wherein a portion of the first pose is compared to the pose parameters for the authentication gesture using a distance metric for poses, wherein a pose includes positions of a set of human joints;
• authenticate the human based on a match between the first pose and the authentication gesture to enable gesture commands from the human;
• input at least a portion of a second image captured using the image sensor to the first machine learning model to obtain a second pose of the human;
• compare the second pose to pose parameters for a follow gesture;
• commence a follow mode based on a match between the second pose and the follow gesture occurring after authentication of the human;
• and
• control the vehicle to follow the human responsive to being in the follow mode.

A non-transitory computer-readable storage medium storing executable instructions that, when executed by a processor, facilitate performance of operations comprising:

• detecting a human in a first image captured using an image sensor connected to a vehicle;
• inputting at least a portion of the first image to a first machine learning model to obtain a first pose of the human, wherein the first pose includes a position of a wrist, an elbow, a shoulder, a waist, a knee, an ankle, or a knuckle;
• comparing the first pose to pose parameters for an authentication gesture, wherein a portion of the first pose is compared to the pose parameters for the authentication gesture using a distance metric for poses, wherein a pose includes positions of a set of human joints;
• authenticating the human based on a match between the first pose and the authentication gesture to enable gesture commands from the human;
• inputting at least a portion of a second image captured using the image sensor to the first machine learning model to obtain a second pose of the human;
• comparing the second pose to pose parameters for a follow gesture;
• commencing a follow mode based on a match between the second pose and the follow gesture occurring after authentication of the human;
• and
• controlling the vehicle to follow the human responsive to being in the follow mode.