package agent

Get desktop application:
View/edit binary Protocol Buffers messages

• rpc Init (InitRequest, InitResponse)

  agent.proto:21

  Initialize MJPC Agent.

  message InitRequest

  agent.proto:67

  • optional string task_id = 1

  • optional MjModel model = 2

  • optional float real_time_speed = 3

    only used on when there is asynchronous planning

  message InitResponse

  agent.proto:73

  (message has no fields)

• rpc GetState (GetStateRequest, GetStateResponse)

  agent.proto:23

  Get the simulation state.

  message GetStateRequest

  agent.proto:85

  (message has no fields)

  message GetStateResponse

  agent.proto:86

  • optional State state = 1

• rpc SetState (SetStateRequest, SetStateResponse)

  agent.proto:25

  Set state of the MJPC Agent.

  message SetStateRequest

  agent.proto:90

  • optional State state = 1

  message SetStateResponse

  agent.proto:93

  (message has no fields)

• rpc GetAction (GetActionRequest, GetActionResponse)

  agent.proto:27

  Get the current action from the Agent.

  message GetActionRequest

  agent.proto:95

  • optional float time = 1

    The time at which the plan should be evaluated, in seconds. If not set, the current time (from the last call to SetState) will be used. Note: for feedback controllers like iLQG, it's probably wrong to request the action in a future time, and instead the state should be set before calling GetAction.

  • optional float averaging_duration = 2

    If not zero, physics will be rolled out for the given duration (in seconds) and actions will be averaged over that period. During the rollout, the task's Transition will not be called.

  • optional bool nominal_action = 3

    For planners that use feedback terms (iLQG), if true, return the nominal action for the given time rather than applying feedback terms on the current state. For the sampling planner this has no effect.

  message GetActionResponse

  agent.proto:114

  • repeated float action = 1

• rpc PlannerStep (PlannerStepRequest, PlannerStepResponse)

  agent.proto:29

  Compute one plan step.

  message PlannerStepRequest

  agent.proto:139

  (message has no fields)

  message PlannerStepResponse

  agent.proto:140

  (message has no fields)

• rpc Step (StepRequest, StepResponse)

  agent.proto:31

  Step physics once, using actions from the planner.

  message StepRequest

  agent.proto:142

  • bool use_previous_policy = 1

    if true, the policy from before the last call to PlanIteration will be used. This is useful for simulating planning delay.

  message StepResponse

  agent.proto:147

  (message has no fields)

• rpc Reset (ResetRequest, ResetResponse)

  agent.proto:33

  Reset the Agent.

  message ResetRequest

  agent.proto:149

  (message has no fields)

  message ResetResponse

  agent.proto:150

  (message has no fields)

• rpc SetTaskParameters (SetTaskParametersRequest, SetTaskParametersResponse)

  agent.proto:35

  Set a task parameters.

  message SetTaskParametersRequest

  agent.proto:161

  • map<string, TaskParameterValue> parameters = 1

    map from parameter name to desired value

  message SetTaskParametersResponse

  agent.proto:165

  (message has no fields)

• rpc GetTaskParameters (GetTaskParametersRequest, GetTaskParametersResponse)

  agent.proto:38

  Get a task parameters.

  message GetTaskParametersRequest

  agent.proto:167

  (message has no fields)

  message GetTaskParametersResponse

  agent.proto:169

  • map<string, TaskParameterValue> parameters = 1

• rpc SetCostWeights (SetCostWeightsRequest, SetCostWeightsResponse)

  agent.proto:41

  Set cost weights.

  message SetCostWeightsRequest

  agent.proto:173

  • bool reset_to_defaults = 1

  • map<string, double> cost_weights = 2

    cost weights by name

  message SetCostWeightsResponse

  agent.proto:178

  (message has no fields)

• rpc GetResiduals (GetResidualsRequest, GetResidualsResponse)

  agent.proto:43

  Get cost term residuals.

  message GetResidualsRequest

  agent.proto:118

  (message has no fields)

  message GetResidualsResponse

  agent.proto:124

  • map<string, Residual> values = 1

• rpc GetCostValuesAndWeights (GetCostValuesAndWeightsRequest, GetCostValuesAndWeightsResponse)

  agent.proto:45

  Get cost term values.

  message GetCostValuesAndWeightsRequest

  agent.proto:128

  (message has no fields)

  message GetCostValuesAndWeightsResponse

  agent.proto:135

  • map<string, ValueAndWeight> values_weights = 1

• rpc SetMode (SetModeRequest, SetModeResponse)

  agent.proto:48

  Set mode.

  message SetModeRequest

  agent.proto:186

  • string mode = 1

  message SetModeResponse

  agent.proto:190

  (message has no fields)

• rpc GetMode (GetModeRequest, GetModeResponse)

  agent.proto:50

  Get mode.

  message GetModeRequest

  agent.proto:180

  (message has no fields)

  message GetModeResponse

  agent.proto:182

  • string mode = 1

• rpc GetAllModes (GetAllModesRequest, GetAllModesResponse)

  agent.proto:52

  Get all modes.

  message GetAllModesRequest

  agent.proto:192

  (message has no fields)

  message GetAllModesResponse

  agent.proto:194

  • repeated string mode_names = 1

• rpc GetBestTrajectory (GetBestTrajectoryRequest, GetBestTrajectoryResponse)

  agent.proto:55

  Get best trajectory (states, actions, times).

  message GetBestTrajectoryRequest

  agent.proto:198

  (message has no fields)

  message GetBestTrajectoryResponse

  agent.proto:200

  • repeated double states = 1

  • repeated double actions = 2

  • repeated double times = 3

  • int32 steps = 4

• rpc SetAnything (SetAnythingRequest, SetAnythingResponse)

  agent.proto:59

  A single method that can set many of the inputs.

  message SetAnythingRequest

  agent.proto:212

  • optional State state = 1

  • map<string, TaskParameterValue> parameters = 2

    map from parameter name to desired value

  • map<string, double> cost_weights = 3

    cost weights by name

  • string mode = 4

  • map<string, Pose> mocap = 5

    set the positions of mocap bodies by name. If `state` is set too, mocap positions will be set after the state is set.

  message SetAnythingResponse

  agent.proto:226

  (message has no fields)

Used in: InitRequest

• optional bytes mjb = 1

• optional string xml = 2

Used in: SetAnythingRequest

• repeated double pos = 1

• repeated double quat = 2

Used in: GetResidualsResponse

• repeated double values = 1

Used in: GetStateResponse, SetAnythingRequest, SetStateRequest

• optional double time = 1

• repeated double qpos = 2

• repeated double qvel = 3

• repeated double act = 4

• repeated double mocap_pos = 5

• repeated double mocap_quat = 6

• repeated double userdata = 7

Used in: GetTaskParametersResponse, SetAnythingRequest, SetTaskParametersRequest

• oneof value

  • double numeric = 1

    used for normal numeric task parameters

  • string selection = 2

    used for values that come from drop down selections

Used in: GetCostValuesAndWeightsResponse

• double value = 1

• double weight = 2